{"models": [{"id": "claude-sonnet-5", "provider": "anthropic", "display_name": "Claude Sonnet 5", "status": "active", "released_at": "2026-06-01", "context_window": 1000000, "cost": {"input_per_1m": 2.0, "output_per_1m": 10.0, "cached_input_per_1m": 0.2, "cache_write_per_1m": 2.5}, "features": ["tool_use", "prompt_caching", "thinking", "vision"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "structured_output", "vision"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "long_context", "structured_output", "general_purpose"], "bad_for": ["latency_critical"], "latency": "medium"}, "aedelgard_note": "Strong cost/capability ratio for most agent tasks at $2/$10. For new work, claude-sonnet-5-5 is the newer Sonnet. [anthropic.com/pricing 2026-09-04]", "price_confirmed_at": "2026-10-05", "_auto_classified": false, "caching": "automatic", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 5.0}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "amazon.nova-pro-v1:0", "provider": "bedrock", "display_name": "Amazon Nova Pro", "status": "active", "released_at": "2024-12-01", "context_window": 300000, "cost": {"input_per_1m": 0.8, "output_per_1m": 3.2, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "vision", "speed"], "cost_efficiency": "high", "classification": {"good_for": ["ingestion", "summarisation", "long_context", "aws_native"], "bad_for": ["complex_reasoning"], "latency": "medium"}, "aedelgard_note": "Best Bedrock option for AWS-native workloads \u2014 IAM auth, no key management. 300k context at low cost.", "price_confirmed_at": "2026-07-29", "caching": "none", "_auto_classified": false, "modality": {"input": ["text", "image", "video", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 10000, "source": "models.dev"}, "purpose": "text"}, {"id": "amazon.nova-lite-v1:0", "provider": "bedrock", "display_name": "Amazon Nova Lite", "status": "active", "released_at": "2024-12-01", "context_window": 300000, "cost": {"input_per_1m": 0.06, "output_per_1m": 0.24, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "fast", "capability_profile": ["speed", "long_context"], "cost_efficiency": "high", "classification": {"good_for": ["high_volume_ingestion", "cheap_summarisation", "aws_native"], "bad_for": ["complex_reasoning", "structured_output"], "latency": "low"}, "aedelgard_note": "Extremely cheap, long context. Good for bulk summarisation where the quality ceiling matters less. amazon.nova-2-lite-v1:0 is its successor.", "price_confirmed_at": "2026-07-29", "caching": "none", "_auto_classified": false, "modality": {"input": ["text", "image", "video", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 10000, "source": "models.dev"}, "purpose": "text"}, {"id": "amazon.nova-micro-v1:0", "provider": "bedrock", "display_name": "Amazon Nova Micro", "status": "active", "released_at": "2024-12-01", "context_window": 300000, "cost": {"input_per_1m": 0.035, "output_per_1m": 0.14, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": [], "tier": "micro", "capability_profile": ["speed"], "cost_efficiency": "high", "classification": {"good_for": ["classification", "triage", "ultra_cheap_extraction"], "bad_for": ["tool_use", "long_context", "reasoning"], "latency": "very_low"}, "aedelgard_note": "Cheapest model in the catalogue. Text-only. Use only for classification/triage where quality is secondary to volume.", "price_confirmed_at": "2026-07-29", "caching": "none", "_auto_classified": false, "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 10000, "source": "models.dev"}, "purpose": "text"}, {"id": "gemini-2.5-pro", "provider": "google", "display_name": "Gemini 2.5 Pro", "status": "active", "released_at": "2025-05-01", "context_window": 1048576, "cost": {"input_per_1m": 1.25, "output_per_1m": 10.0, "cached_input_per_1m": 0.31, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context", "thinking", "multimodal"], "tier": "legacy", "capability_profile": ["long_context", "reasoning", "multimodal", "tool_use"], "cost_efficiency": "medium", "classification": {"good_for": ["very_long_context", "multimodal", "reasoning"], "bad_for": ["latency_critical"], "latency": "high"}, "aedelgard_note": "Prior-generation Pro, kept active for bodies already on it. New work \u2192 gemini-3.1-pro-preview ($2/$12 vs $1.25/$10, but a generation newer). [ai.google.dev pricing 2026-09-04]", "price_confirmed_at": "2026-07-29", "caching": "explicit_unwired", "_auto_classified": false, "modality": {"input": ["text", "image", "audio", "video", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions"], "max_output_tokens": 65535, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.625, "output_per_1m": 5.0}, "priority": {"input_per_1m": 2.25, "output_per_1m": 18.0}}, "context_step": {"above_tokens": 200000, "input_per_1m": 2.5, "output_per_1m": 15.0}, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-2.5-flash", "provider": "google", "display_name": "Gemini 2.5 Flash", "status": "active", "released_at": "2025-05-01", "context_window": 1048576, "cost": {"input_per_1m": 0.3, "output_per_1m": 2.5, "cached_input_per_1m": 0.075, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context", "thinking"], "tier": "legacy", "capability_profile": ["speed", "long_context", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": ["long_context_cheap", "high_volume", "multimodal_cheap"], "bad_for": ["frontier_reasoning"], "latency": "medium"}, "aedelgard_note": "Prior-generation Flash, kept active for bodies already on it. New work \u2192 gemini-3.8-flash (same 1M context, newer, comparable price). [ai.google.dev pricing 2026-09-04]", "price_confirmed_at": "2026-07-29", "caching": "explicit_unwired", "_auto_classified": false, "modality": {"input": ["text", "image", "audio", "video", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65535, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.15, "output_per_1m": 1.25}, "priority": {"input_per_1m": 0.54, "output_per_1m": 4.5}}, "context_step": null, "reasoning_output_per_1m": 2.5, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-fable-5", "provider": "anthropic", "display_name": "Claude Fable 5", "status": "active", "released_at": "2026-07-29", "context_window": 1000000, "cost": {"input_per_1m": 10.0, "output_per_1m": 50.0, "cached_input_per_1m": 1.0, "cache_write_per_1m": 12.5}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "general_purpose", "long_context"], "bad_for": ["frontier_reasoning"], "latency": "medium"}, "aedelgard_note": "Prior Fable. Same $10/$50 as Fable 5.1 but cache reads cost 4\u00d7 more ($1.00 vs $0.25) \u2014 prefer claude-fable-5-1 for anything new. [anthropic.com/pricing 2026-09-04]", "price_confirmed_at": "2026-08-03", "_auto_classified": false, "good_for": ["tool_cascades", "general_purpose", "long_context"], "bad_for": ["frontier_reasoning"], "latency": "medium", "caching": "automatic", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 5.0, "output_per_1m": 25.0}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-opus-4-8", "provider": "anthropic", "display_name": "Claude Opus 4.8", "status": "active", "released_at": "2026-07-29", "context_window": 1000000, "cost": {"input_per_1m": 5.0, "output_per_1m": 25.0, "cached_input_per_1m": 0.5, "cache_write_per_1m": 6.25}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "extended_thinking"], "cost_efficiency": "low", "classification": {"good_for": ["complex_reasoning", "extended_thinking"], "bad_for": ["high_volume", "cost_sensitive"], "latency": "high"}, "aedelgard_note": "Prior Opus generation. Prefer claude-opus-5-5 for new work (cheaper and newer).", "price_confirmed_at": "2026-08-03", "_auto_classified": false, "good_for": ["complex_reasoning", "extended_thinking"], "bad_for": ["high_volume", "cost_sensitive"], "latency": "high", "caching": "automatic", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 2.5, "output_per_1m": 12.5}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-sonnet-4-6", "provider": "anthropic", "display_name": "Claude Sonnet 4.6", "status": "active", "released_at": "2026-07-29", "context_window": 1000000, "cost": {"input_per_1m": 3.0, "output_per_1m": 15.0, "cached_input_per_1m": 0.3, "cache_write_per_1m": 3.75}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "general_purpose"], "bad_for": ["frontier_reasoning"], "latency": "medium"}, "aedelgard_note": "Prior Sonnet at $3/$15 \u2014 now MORE expensive than Sonnet 5 ($2/$10). No reason to pick it for new work; prefer claude-sonnet-5. [anthropic.com/pricing 2026-09-04]", "price_confirmed_at": "2026-08-03", "_auto_classified": false, "good_for": ["tool_cascades", "general_purpose"], "bad_for": ["frontier_reasoning"], "latency": "medium", "caching": "automatic", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.5, "output_per_1m": 7.5}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-haiku-4-5-20251001", "provider": "anthropic", "display_name": "Claude Haiku 4.5", "status": "active", "released_at": "2026-07-29", "context_window": 200000, "cost": {"input_per_1m": 1.0, "output_per_1m": 5.0, "cached_input_per_1m": 0.1, "cache_write_per_1m": 1.25}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "micro", "capability_profile": ["speed", "tool_use", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["summarisation", "ingestion", "triage", "high_volume"], "bad_for": ["complex_reasoning"], "latency": "low"}, "aedelgard_note": "Anthropic cheap/fast tier \u2014 optimal for compaction, trims, and triage. Price corrected 2026-10-05 to $1/$5 (was mis-recorded as $0.80/$4). [vendor pricing page + LiteLLM, cross-checked 2026-10-05]", "price_confirmed_at": "2026-10-05", "_auto_classified": false, "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 64000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.5, "output_per_1m": 2.5}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gpt-5.6-terra", "provider": "openai", "display_name": "gpt-5.6-terra", "status": "active", "released_at": "2026-07-29", "context_window": 1050000, "cost": {"input_per_1m": 2.0, "output_per_1m": 12.0, "cached_input_per_1m": 0.2, "cache_write_per_1m": 2.5}, "features": ["tool_use", "vision", "prompt_caching", "structured_output", "reasoning"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "vision", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "general_purpose"], "bad_for": ["latency_critical"], "latency": "medium"}, "aedelgard_note": "OpenAI's workhorse tier of the 5.6 line, the Sonnet-equivalent. For new work, gpt-6-sol is newer at a similar price.", "price_confirmed_at": "2026-09-06", "_auto_classified": false, "caching": "automatic", "reasoning_effort": ["minimal", "low", "medium", "high"], "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/batch", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 6.0}, "priority": {"input_per_1m": 4.0, "output_per_1m": 24.0}}, "context_step": {"above_tokens": 272000, "input_per_1m": 4.0, "output_per_1m": 18.0}, "reasoning_output_per_1m": null, "effort_supported": ["none", "xhigh"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-flash-latest", "provider": "google", "display_name": "Gemini Flash Latest", "status": "active", "released_at": "2026-07-29", "context_window": 1048576, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "fast", "capability_profile": ["speed", "long_context", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": ["ingestion", "summarisation", "speed"], "bad_for": ["complex_reasoning"], "latency": "low"}, "aedelgard_note": "Google's floating alias to their current-best Flash model. Same caveat as gemini-pro-latest \u2014 floats, prefer gemini-2.5-flash if you want stability.", "price_confirmed_at": null, "_auto_classified": false, "caching": "explicit_unwired", "reasoning_effort": null, "modality": {"input": ["text", "image", "video", "audio", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.375, "output_per_1m": 1.875}, "priority": {"input_per_1m": 1.35, "output_per_1m": 6.75}}, "context_step": null, "reasoning_output_per_1m": 3.75, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-flash-lite-latest", "provider": "google", "display_name": "Gemini Flash-Lite Latest", "status": "active", "released_at": "2026-07-29", "context_window": 1048576, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "micro", "capability_profile": ["speed", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Google's floating alias to their current-best Flash-Lite (today: 3.5-flash-lite). Same caveat as the other -latest aliases: the brain under you can change without a release.", "price_confirmed_at": null, "_auto_classified": false, "caching": "explicit_unwired", "modality": {"input": ["text", "image", "video", "audio", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.15, "output_per_1m": 1.25}, "priority": {"input_per_1m": 0.54, "output_per_1m": 4.5}}, "context_step": null, "reasoning_output_per_1m": 2.5, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-pro-latest", "provider": "google", "display_name": "Gemini Pro Latest", "status": "active", "released_at": "2026-07-29", "context_window": 1048576, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "multimodal", "vision"], "cost_efficiency": "medium", "classification": {"good_for": ["complex_reasoning", "long_context", "vision"], "bad_for": ["cost_sensitive"], "latency": "high"}, "aedelgard_note": "Google's floating alias to their current-best Pro model \u2014 avoids preview churn. Prefer gemini-2.5-pro if you need a pinned, stable id.", "price_confirmed_at": null, "_auto_classified": false, "caching": "explicit_unwired", "reasoning_effort": null, "modality": {"input": ["text", "image", "video", "audio"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 6.0}, "priority": {"input_per_1m": 3.6, "output_per_1m": 21.6}}, "context_step": {"above_tokens": 200000, "input_per_1m": 4.0, "output_per_1m": 18.0}, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-3.1-pro-preview", "provider": "google", "display_name": "Gemini 3.1 Pro (preview)", "status": "active", "released_at": "2026-07-29", "context_window": 1048576, "cost": {"input_per_1m": 2.0, "output_per_1m": 12.0, "cached_input_per_1m": 0.2, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "multimodal", "structured_output"], "cost_efficiency": "medium", "classification": {"good_for": ["long_context", "multimodal_understanding", "generation", "coding"], "bad_for": ["latency_critical"], "latency": "medium"}, "aedelgard_note": "Google's 3rd-gen Pro, still labelled preview. $2/$12 (\u2264200k prompts; $4/$18 above). Prefer over gemini-2.5-pro for new work; watch for the GA id. [ai.google.dev pricing 2026-09-04]", "price_confirmed_at": "2026-09-04", "_auto_classified": false, "caching": "explicit_unwired", "reasoning_effort": null, "modality": {"input": ["text", "image", "video", "audio", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 6.0}, "priority": {"input_per_1m": 3.6, "output_per_1m": 21.6}}, "context_step": {"above_tokens": 200000, "input_per_1m": 4.0, "output_per_1m": 18.0}, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gemini-3.5-flash-lite", "provider": "google", "display_name": "Gemini 3.5 Flash-Lite", "status": "active", "released_at": "2026-07-29", "context_window": 1048576, "cost": {"input_per_1m": 0.3, "output_per_1m": 2.5, "cached_input_per_1m": 0.03, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "micro", "capability_profile": ["speed", "long_context", "multimodal"], "cost_efficiency": "high", "classification": {"good_for": ["ingestion", "classification", "translation", "high_volume"], "bad_for": ["deep_reasoning"], "latency": "low"}, "aedelgard_note": "Google's current Lite \u2014 1M context at $0.30/$2.50. The Nova-Lite-class option when you want Google. [ai.google.dev pricing 2026-09-04]", "price_confirmed_at": "2026-09-04", "_auto_classified": false, "caching": "explicit_unwired", "reasoning_effort": null, "modality": {"input": ["text", "image", "video", "audio", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.15, "output_per_1m": 1.25}, "priority": {"input_per_1m": 0.54, "output_per_1m": 4.5}}, "context_step": null, "reasoning_output_per_1m": 2.5, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "anthropic.claude-opus-5", "provider": "bedrock", "display_name": "Claude Opus 5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning"], "cost_efficiency": "low", "classification": {"good_for": ["complex_reasoning"], "bad_for": ["cost_sensitive"], "latency": "high"}, "aedelgard_note": "Bedrock-hosted claude-opus-5. Use when IAM auth is required.", "price_confirmed_at": null, "_auto_classified": false, "good_for": ["complex_reasoning"], "bad_for": ["cost_sensitive"], "latency": "high", "caching": "none", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "qwen.qwen3-235b-a22b-2507-v1:0", "provider": "bedrock", "display_name": "Qwen3 235B A22B 2507", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "medium", "classification": {"good_for": ["reasoning", "tool_use"], "bad_for": [], "latency": "high"}, "aedelgard_note": "Qwen3 235B MoE \u2014 Alibaba's frontier model via Bedrock. Strong reasoning.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "moonshotai.kimi-k2.5", "provider": "bedrock", "display_name": "Kimi K2.5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["long_context", "tool_use"], "cost_efficiency": "medium", "classification": {"good_for": ["long_context", "tool_use"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Kimi K2.5 \u2014 Moonshot AI's long-context model via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 16000, "source": "models.dev"}, "purpose": "text"}, {"id": "openai.gpt-oss-120b-1:0", "provider": "bedrock", "display_name": "gpt-oss-120b", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "reasoning"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "medium"}, "aedelgard_note": "OpenAI open-weight 120B MoE via Bedrock. Solid reasoning at open-weight prices; no vendor lock. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "anthropic.claude-fable-5", "provider": "bedrock", "display_name": "Claude Fable 5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": ["general_purpose"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted claude-fable-5. Use when IAM auth is required.", "price_confirmed_at": null, "_auto_classified": false, "good_for": ["general_purpose"], "bad_for": [], "latency": "medium", "caching": "none", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek.v3.2", "provider": "bedrock", "display_name": "DeepSeek V3.2", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "speed"], "cost_efficiency": "high", "classification": {"good_for": ["tool_use", "high_volume"], "bad_for": ["long_context"], "latency": "medium"}, "aedelgard_note": "DeepSeek V3.2 on Bedrock \u2014 latest DeepSeek available via IAM.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 81920, "source": "models.dev"}, "purpose": "text"}, {"id": "anthropic.claude-sonnet-5", "provider": "bedrock", "display_name": "Claude Sonnet 5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "general_purpose"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted claude-sonnet-5. Use when AWS IAM auth is required instead of an Anthropic key.", "price_confirmed_at": null, "_auto_classified": false, "good_for": ["tool_cascades", "general_purpose"], "bad_for": [], "latency": "medium", "caching": "none", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "qwen.qwen3-coder-480b-a35b-v1:0", "provider": "bedrock", "display_name": "Qwen3 Coder 480B A35B Instruct", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "reasoning", "long_context"], "cost_efficiency": "medium", "classification": {"good_for": ["coding", "tool_cascades"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Qwen3-Coder 480B MoE \u2014 coding-specialised frontier via Bedrock. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 65536, "source": "models.dev"}, "purpose": "text"}, {"id": "amazon.nova-2-lite-v1:0", "provider": "bedrock", "display_name": "Nova 2 Lite", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast", "capability_profile": ["speed", "long_context"], "cost_efficiency": "high", "classification": {"good_for": ["ingestion", "summarisation", "high_volume"], "bad_for": ["complex_reasoning"], "latency": "low"}, "aedelgard_note": "Nova 2 Lite \u2014 successor to Nova Lite. Prefer for new Bedrock ingestion workloads.", "price_confirmed_at": null, "_auto_classified": false, "good_for": ["ingestion", "summarisation", "high_volume"], "bad_for": ["complex_reasoning"], "latency": "low", "caching": "none", "modality": {"input": ["text", "image", "video", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 65535, "source": "models.dev"}, "purpose": "text"}, {"id": "minimax.minimax-m2.5", "provider": "bedrock", "display_name": "MiniMax M2.5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["long_context", "multimodal"], "cost_efficiency": "medium", "classification": {"good_for": ["long_context", "multimodal"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "MiniMax M2.5 \u2014 long-context multimodal model via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 98304, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek.v3-v1:0", "provider": "bedrock", "display_name": "DeepSeek-V3.1", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "speed", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_use", "high_volume"], "bad_for": ["long_context"], "latency": "medium"}, "aedelgard_note": "DeepSeek V3 on Bedrock \u2014 IAM auth, no key needed. Exceptional value.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 81920, "source": "models.dev"}, "purpose": "text"}, {"id": "qwen.qwen3-coder-30b-a3b-v1:0", "provider": "bedrock", "display_name": "Qwen3-Coder-30B-A3B-Instruct", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast", "capability_profile": ["tool_use", "speed"], "cost_efficiency": "high", "classification": {"good_for": ["coding"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Qwen3-Coder 30B (3B active) \u2014 fast coding model via Bedrock. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "nvidia.nemotron-super-3-120b", "provider": "bedrock", "display_name": "NVIDIA Nemotron 3 Super 120B A12B", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use"], "cost_efficiency": "medium", "classification": {"good_for": ["reasoning", "scientific"], "bad_for": ["latency_critical"], "latency": "high"}, "aedelgard_note": "NVIDIA Nemotron 120B \u2014 strong reasoning, scientifically-oriented via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "qwen.qwen3-32b-v1:0", "provider": "bedrock", "display_name": "Qwen3 32B (dense)", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "reasoning"], "cost_efficiency": "high", "classification": {"good_for": ["tool_use", "reasoning"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Qwen3 32B \u2014 dense model, fast, solid tool_use via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 16384, "source": "models.dev"}, "purpose": "text"}, {"id": "zai.glm-5", "provider": "bedrock", "display_name": "GLM 5", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "reasoning"], "cost_efficiency": "medium", "classification": {"good_for": ["tool_use", "reasoning"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "GLM-5 \u2014 Zhipu AI's frontier model via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "openai.gpt-oss-20b-1:0", "provider": "bedrock", "display_name": "gpt-oss-20b", "status": "active", "released_at": "2026-07-29", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast", "capability_profile": ["speed", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": ["deep_reasoning"], "latency": "medium"}, "aedelgard_note": "OpenAI open-weight 20B via Bedrock \u2014 small, fast, fine for triage. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "openai.gpt-5.6-terra", "provider": "bedrock", "display_name": "GPT-5.6 Terra", "status": "active", "released_at": "2026-08-15", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted gpt-5.6-terra (OpenAI's workhorse tier). Same caveat: no reasoning_effort dial, no caching under Converse. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 4.4, "output_per_1m": 19.8}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-5.6-luna", "provider": "bedrock", "display_name": "GPT-5.6 Luna", "status": "active", "released_at": "2026-08-15", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "micro", "capability_profile": ["speed", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": ["deep_reasoning"], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted gpt-5.6-luna \u2014 cheap/fast tier for ingestion and triage. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 0.44, "output_per_1m": 1.98}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-5.6-sol", "provider": "bedrock", "display_name": "GPT-5.6 Sol", "status": "active", "released_at": "2026-08-15", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "reasoning", "long_context", "structured_output"], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted gpt-5.6-sol. Use when AWS IAM auth is required. Through Bedrock the Aedelgard body sends neither the effort setting nor cache hints; the direct OpenAI row has both. Pricing unconfirmed (no vendor page read), billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 8.8, "output_per_1m": 33.0}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "xai.grok-4.6", "provider": "bedrock", "display_name": "Grok 4.6", "status": "active", "released_at": "2026-08-20", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "reasoning", "long_context"], "cost_efficiency": "medium", "classification": {"good_for": ["reasoning", "tool_cascades"], "bad_for": [], "latency": "medium"}, "aedelgard_note": "xAI Grok 4.6 via Bedrock \u2014 the only xAI route in the catalogue with IAM auth. Pricing unconfirmed (no vendor page read) \u2014 IAM-billed via Bedrock.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 500000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"priority": {"input_per_1m": 3.85, "output_per_1m": 11.55}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-fable-5-1", "provider": "anthropic", "display_name": "Claude Fable 5.1", "status": "active", "released_at": "2026-09-02", "context_window": 1000000, "cost": {"input_per_1m": 10.0, "output_per_1m": 50.0, "cached_input_per_1m": 0.25, "cache_write_per_1m": 12.5}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "structured_output"], "cost_efficiency": "low", "classification": {"good_for": ["long_running_agents", "tool_cascades", "deep_planning"], "bad_for": ["cost_sensitive", "latency_critical"], "latency": "medium"}, "aedelgard_note": "Fable-class for long-horizon agents. Same $10/$50 as Fable 5 but cache reads are 4\u00d7 cheaper ($0.25 vs $1.00), so it is the better Fable for cache-heavy persistent agents; supersedes Fable 5. 5\u00d7 Sonnet 5's price: use only when the work needs Fable-tier judgement. [anthropic.com/pricing 2026-09-04]", "price_confirmed_at": "2026-10-05", "_auto_classified": false, "caching": "automatic", "reasoning_effort": null, "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 5.0, "output_per_1m": 25.0}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "anthropic.claude-fable-5-1", "provider": "bedrock", "display_name": "Claude Fable 5.1 (Bedrock)", "status": "active", "released_at": "2026-09-02", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "structured_output"], "cost_efficiency": "low", "classification": {"good_for": ["long_running_agents", "tool_cascades"], "bad_for": ["cost_sensitive"], "latency": "medium"}, "aedelgard_note": "Bedrock-hosted claude-fable-5-1. Use when AWS IAM auth is required. The Aedelgard body sends no prompt-cache blocks through Bedrock, so the direct Anthropic row is cheaper for cache-heavy agents.", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "gemini-3.8-flash", "provider": "google", "display_name": "Gemini 3.8 Flash", "status": "active", "released_at": "2026-09-03", "context_window": 1048576, "cost": {"input_per_1m": 0.75, "output_per_1m": 3.75, "cached_input_per_1m": 0.075, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "fast", "capability_profile": ["tool_use", "long_context", "speed", "multimodal", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["tool_cascades", "long_context", "coding", "high_throughput"], "bad_for": [], "latency": "low"}, "aedelgard_note": "Google's current Flash (released 2026-09-03): 1M context at $0.75/$3.75 through 2026-12-31, then $1.50/$7.50. Best cost/context ratio in the catalogue for agentic work; replaces 2.5-flash as the Google fast pick. Google offers context caching; the Aedelgard body does not use it yet. [ai.google.dev pricing 2026-09-04]", "price_confirmed_at": "2026-09-04", "_auto_classified": false, "caching": "explicit_unwired", "reasoning_effort": null, "modality": {"input": ["text", "image", "video", "audio", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/completions", "/v1/batch"], "max_output_tokens": 65536, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.375, "output_per_1m": 1.875}, "priority": {"input_per_1m": 1.35, "output_per_1m": 6.75}}, "context_step": null, "reasoning_output_per_1m": 3.75, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gpt-6-astra", "provider": "openai", "display_name": "GPT-6 Astra", "status": "active", "released_at": "2026-09-05", "context_window": 1050000, "cost": {"input_per_1m": 10.0, "output_per_1m": 50.0, "cached_input_per_1m": 1.0, "cache_write_per_1m": 12.5}, "features": ["tool_use", "vision", "prompt_caching", "structured_output", "reasoning"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "vision", "structured_output"], "cost_efficiency": "low", "classification": {"good_for": ["complex_reasoning", "multi_step_planning", "coding", "computer_use", "vision"], "bad_for": ["latency_critical", "cost_sensitive"], "latency": "high"}, "aedelgard_note": "OpenAI's top model (2026-09-05): 'most capable model, built for the hardest end-to-end work'. 1.05M window (922k max input), 128k output, cutoff Apr 30 2026, effort low->max. Price peer of Claude Fable 5.1 ($10/$50): the ceiling, not the default; gpt-6-sol at $2/$10 covers most days. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "automatic", "reasoning_effort": ["low", "medium", "high", "xhigh", "max"], "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/batch", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 5.0, "output_per_1m": 25.0}, "priority": {"input_per_1m": 20.0, "output_per_1m": 100.0}}, "context_step": {"above_tokens": 272000, "input_per_1m": 20.0, "output_per_1m": 75.0}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-6-astra", "provider": "bedrock", "display_name": "GPT-6 Astra", "status": "active", "released_at": "2026-09-09", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Bedrock-hosted alias of gpt-6-astra. Use when IAM auth is required. Pricing is Bedrock's own and is NOT confirmed here \u2014 route to the native provider for cost-critical work. [read 2026-09-23]", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 22.0, "output_per_1m": 82.5}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gpt-image-2.5-flare", "provider": "openai", "display_name": "GPT-Image-2.5 Flare", "status": "active", "released_at": "2026-09-04", "context_window": 128000, "cost": {"input_per_1m": 5.0, "output_per_1m": 30.0, "cached_input_per_1m": 2.0, "cache_write_per_1m": null}, "features": ["image_generation", "image_editing"], "tier": "fast", "capability_profile": ["image_generation", "image_editing"], "cost_efficiency": "medium", "classification": {"good_for": ["image_generation", "everyday_visuals"], "bad_for": ["text_tasks"], "latency": "low"}, "aedelgard_note": "GPT Image 2.5 Flare \u2014 fastest high-quality everyday image generation (2026-09-04). Text in $5/M \u00b7 image in $8/M \u00b7 image out $30/M \u00b7 cached image $2/M. Default snapshot gpt-image-2.5-flare-2026-09-08. Not eligible for text recommendation. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image"], "output": ["image"], "endpoints": ["/v1/images/generations", "/v1/images/edits"], "max_output_tokens": null, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 2.5, "output_per_1m": null}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm"}, "purpose": "image"}, {"id": "gpt-image-2.5-sunburst", "provider": "openai", "display_name": "GPT-Image-2.5 Sunburst", "status": "active", "released_at": "2026-09-04", "context_window": 128000, "cost": {"input_per_1m": 5.0, "output_per_1m": 30.0, "cached_input_per_1m": 2.0, "cache_write_per_1m": null}, "features": ["image_generation", "image_editing"], "tier": "fast", "capability_profile": ["image_generation", "image_editing"], "cost_efficiency": "medium", "classification": {"good_for": ["image_editing", "editing_precision"], "bad_for": ["text_tasks"], "latency": "low"}, "aedelgard_note": "GPT Image 2.5 Sunburst \u2014 image generation/editing where editing precision matters most (2026-09-04). Same rates as Flare: text in $5/M \u00b7 image in $8/M \u00b7 image out $30/M. Default snapshot gpt-image-2.5-sunburst-2026-09-08. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "image"], "output": ["image"], "endpoints": ["/v1/images/generations", "/v1/images/edits"], "max_output_tokens": null, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 2.5, "output_per_1m": null}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm"}, "purpose": "image"}, {"id": "gpt-live-1", "provider": "openai", "display_name": "GPT-Live 1", "status": "active", "released_at": "2026-09-08", "context_window": null, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["voice", "realtime", "audio"], "tier": "fast", "capability_profile": ["realtime", "voice"], "cost_efficiency": "high", "classification": {"good_for": ["voice_agents", "realtime_conversation"], "bad_for": ["text_tasks"], "latency": "low"}, "aedelgard_note": "Full-duplex voice model (2026-09-08): listens and speaks simultaneously and can delegate reasoning/tool use to a backend agent. Billed $0.05/minute (per second); backend model and tools billed separately. NO token pricing \u2014 not eligible for text recommendation. cutoff Jul 31 2025. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "none", "reasoning_effort": null, "modality": {"input": ["text", "audio"], "output": ["text", "audio"], "endpoints": null, "max_output_tokens": null, "source": "litellm+models.dev"}, "purpose": "audio"}, {"id": "moonshotai.kimi-k3", "provider": "bedrock", "display_name": "Kimi K3", "status": "active", "released_at": "2026-09-18", "context_window": 300000, "cost": {"input_per_1m": 3.3, "output_per_1m": 16.5, "cached_input_per_1m": 0.33, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from aws-price-list:bedrock@eu-north-1. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "aws-price-list:bedrock@eu-north-1", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "Qwen/Qwen3-235B-A22B-Instruct-2507", "provider": "nebius", "display_name": "Qwen3 235B A22B Instruct 2507", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.2, "output_per_1m": 0.6, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/Qwen/Qwen3-235B-A22B-Instruct-2507. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/Qwen/Qwen3-235B-A22B-Instruct-2507", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "google/gemma-3-27b-it", "provider": "nebius", "display_name": "gemma 3 27b it", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.3, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/google/gemma-3-27b-it. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/google/gemma-3-27b-it", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 110000, "source": "models.dev"}, "purpose": "text"}, {"id": "openai/gpt-oss-120b", "provider": "nebius", "display_name": "gpt oss 120b", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.6, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/openai/gpt-oss-120b. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/openai/gpt-oss-120b", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "Qwen/Qwen3-30B-A3B-Instruct-2507", "provider": "nebius", "display_name": "Qwen3 30B A3B Instruct 2507", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.3, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/Qwen/Qwen3-30B-A3B-Instruct-2507. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/Qwen/Qwen3-30B-A3B-Instruct-2507", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "NousResearch/Hermes-4-405B", "provider": "nebius", "display_name": "Hermes 4 405B", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 1.0, "output_per_1m": 3.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/NousResearch/Hermes-4-405B. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/NousResearch/Hermes-4-405B", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B", "provider": "nebius", "display_name": "NVIDIA Nemotron 3 Nano 30B A3B", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.06, "output_per_1m": 0.24, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["speed", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "zai-org/GLM-5.1", "provider": "nebius", "display_name": "GLM 5.1", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 1.4, "output_per_1m": 4.4, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/zai-org/GLM-5.1. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/zai-org/GLM-5.1", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 202752, "source": "models.dev"}, "purpose": "text"}, {"id": "nvidia/Nemotron-3_5-Lightning", "provider": "nebius", "display_name": "Nemotron 3_5 Lightning", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.06, "output_per_1m": 0.24, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["speed", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/nvidia/Nemotron-3_5-Lightning. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/nvidia/Nemotron-3_5-Lightning", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1048576, "source": "models.dev"}, "purpose": "text"}, {"id": "openbmb/MiniCPM-V-4_5", "provider": "nebius", "display_name": "MiniCPM V 4_5", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.658, "output_per_1m": 1.11, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast-flagship from family/size; price from litellm:nebius/openbmb/MiniCPM-V-4_5. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/openbmb/MiniCPM-V-4_5", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 32000, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek-ai/DeepSeek-V4-Flash-0731", "provider": "nebius", "display_name": "DeepSeek V4 Flash (Nebius EU)", "status": "active", "released_at": "2026-09-19", "context_window": 1000000, "cost": {"input_per_1m": 0.14, "output_per_1m": 0.28, "cached_input_per_1m": 0.035, "cache_write_per_1m": 0.14}, "features": ["tool_use", "speed", "long_context", "streaming"], "tier": "fast", "capability_profile": ["speed", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "DeepSeek 304B fast MoE hosted in M\u00e4nts\u00e4l\u00e4, Finland via Nebius Token Factory. 1M context.", "price_confirmed_at": "2026-09-19", "_auto_classified": false, "caching": "automatic", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1024000, "source": "models.dev"}, "purpose": "text"}, {"id": "MiniMaxAI/MiniMax-M3", "provider": "nebius", "display_name": "MiniMax M3", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 1.2, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast-flagship from family/size; price from litellm:nebius/MiniMaxAI/MiniMax-M3. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/MiniMaxAI/MiniMax-M3", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1048576, "source": "models.dev"}, "purpose": "text"}, {"id": "moonshotai/Kimi-K2.7-Code", "provider": "nebius", "display_name": "Kimi K2.7 Code", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.95, "output_per_1m": 4.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/moonshotai/Kimi-K2.7-Code. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/moonshotai/Kimi-K2.7-Code", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "nvidia/Nemotron-3-Ultra-550b-a55b", "provider": "nebius", "display_name": "Nemotron 3 Ultra 550b a55b", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 1.0, "output_per_1m": 3.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/nvidia/Nemotron-3-Ultra-550b-a55b. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/nvidia/Nemotron-3-Ultra-550b-a55b", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1048576, "source": "models.dev"}, "purpose": "text"}, {"id": "moonshotai/Kimi-K2.6", "provider": "nebius", "display_name": "Kimi K2.6", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.95, "output_per_1m": 4.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/moonshotai/Kimi-K2.6. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/moonshotai/Kimi-K2.6", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text", "image", "video"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "nvidia/nemotron-3-super-120b-a12b", "provider": "nebius", "display_name": "nemotron 3 super 120b a12b", "status": "active", "released_at": "2026-09-19", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 0.9, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/nvidia/nemotron-3-super-120b-a12b. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/nvidia/nemotron-3-super-120b-a12b", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "Qwen/Qwen3.5-397B-A17B", "provider": "nebius", "display_name": "Qwen3.5 397B A17B", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 0.6, "output_per_1m": 3.6, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/Qwen/Qwen3.5-397B-A17B. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/Qwen/Qwen3.5-397B-A17B", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek-ai/DeepSeek-V4-Pro", "provider": "nebius", "display_name": "DeepSeek V4 Pro (Nebius EU)", "status": "active", "released_at": "2026-09-19", "context_window": 1000000, "cost": {"input_per_1m": 0.55, "output_per_1m": 2.19, "cached_input_per_1m": 0.14, "cache_write_per_1m": 0.55}, "features": ["tool_use", "reasoning", "long_context", "streaming"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "DeepSeek 862B MoE flagship hosted in M\u00e4nts\u00e4l\u00e4, Finland via Nebius Token Factory. Direct EU jurisdiction.", "price_confirmed_at": "2026-09-19", "_auto_classified": false, "caching": "automatic", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1048576, "source": "models.dev"}, "purpose": "text"}, {"id": "zai-org/GLM-5.2", "provider": "nebius", "display_name": "GLM 5.2", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 1.4, "output_per_1m": 4.4, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/zai-org/GLM-5.2. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/zai-org/GLM-5.2", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1048576, "source": "models.dev"}, "purpose": "text"}, {"id": "moonshotai/Kimi-K3", "provider": "nebius", "display_name": "Kimi K3", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 3.0, "output_per_1m": 15.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/moonshotai/Kimi-K3. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/moonshotai/Kimi-K3", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1024000, "source": "models.dev"}, "purpose": "text"}, {"id": "zai-org/GLM-5.3-Flash", "provider": "nebius", "display_name": "GLM 5.3 Flash", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.5, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast", "capability_profile": ["speed", "tool_use"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=fast from family/size; price from litellm:nebius/zai-org/GLM-5.3-Flash. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/zai-org/GLM-5.3-Flash", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1024000, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek-ai/DeepSeek-V4.1-Flash", "provider": "nebius", "display_name": "DeepSeek V4.1 Flash (Nebius EU)", "status": "active", "released_at": "2026-09-19", "context_window": 1000000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.3, "cached_input_per_1m": 0.038, "cache_write_per_1m": 0.15}, "features": ["tool_use", "vision", "speed", "long_context", "streaming"], "tier": "fast", "capability_profile": ["speed", "tool_use", "multimodal", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "DeepSeek V4.1 multimodal Flash hosted in Finland via Nebius Token Factory. 1M context.", "price_confirmed_at": "2026-09-19", "_auto_classified": false, "caching": "automatic", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 384000, "source": "models.dev"}, "purpose": "text"}, {"id": "zai-org/GLM-5.3", "provider": "nebius", "display_name": "GLM 5.3", "status": "active", "released_at": "2026-09-19", "context_window": 131072, "cost": {"input_per_1m": 1.4, "output_per_1m": 4.4, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:nebius/zai-org/GLM-5.3. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:nebius/zai-org/GLM-5.3", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 1024000, "source": "models.dev"}, "purpose": "text"}, {"id": "deepseek-ai/DeepSeek-V4-Pro-0813", "provider": "nebius", "display_name": "DeepSeek V4 Pro 0813 (Nebius EU)", "status": "active", "released_at": "2026-09-19", "context_window": 1000000, "cost": {"input_per_1m": 0.55, "output_per_1m": 2.19, "cached_input_per_1m": 0.14, "cache_write_per_1m": 0.55}, "features": ["tool_use", "reasoning", "long_context", "streaming"], "tier": "flagship", "capability_profile": ["reasoning", "tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "DeepSeek 862B MoE (checkpoint 0813) hosted in Finland via Nebius Token Factory.", "price_confirmed_at": "2026-09-19", "_auto_classified": false, "caching": "automatic", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 979000, "source": "models.dev"}, "reactivated_at": "2026-10-10", "purpose": "text"}, {"id": "claude-opus-5-5", "provider": "anthropic", "display_name": "Claude Opus 5.5", "status": "active", "released_at": "2026-09-21", "context_window": 1000000, "cost": {"input_per_1m": 4.0, "output_per_1m": 20.0, "cached_input_per_1m": 0.2, "cache_write_per_1m": 5.0}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["general_workloads", "complex_reasoning", "agentic_coding", "long_context"], "bad_for": ["latency_critical"], "latency": "medium"}, "aedelgard_note": "Anthropic's recommended default (2026-09-21): 'start with Claude Opus 5.5 for most workloads'. Same $4/$20 as Opus 5 but a Jun-2026 cutoff and cache reads at 0.05x ($0.20 vs $0.50), the cheapest flagship per cached token in the catalogue. Fable 5.1 remains the ceiling for long-horizon agentic work. 5m cache write $5 / 1h $8. [read 2026-09-23]", "price_confirmed_at": "2026-10-05", "_auto_classified": false, "caching": "automatic", "reasoning_effort": null, "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 2.0, "output_per_1m": 10.0}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gpt-6-sol", "provider": "openai", "display_name": "GPT-6 Sol", "status": "active", "released_at": "2026-09-14", "context_window": 1050000, "cost": {"input_per_1m": 2.0, "output_per_1m": 10.0, "cached_input_per_1m": 0.2, "cache_write_per_1m": 2.5}, "features": ["tool_use", "vision", "prompt_caching", "structured_output", "reasoning"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "reasoning", "vision", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["agentic_coding", "tool_cascades", "long_context", "vision", "complex_reasoning"], "bad_for": ["latency_critical"], "latency": "medium"}, "aedelgard_note": "GPT-6 workhorse flagship (2026-09-14), built for complex coding and agentic workflows. $2/$10, half of GPT-5.6 Sol and level with Claude Sonnet 5, at a 1.05M window (922k max input) / 128k output, cutoff Apr 20 2026. Effort none/low/medium(default)/high/xhigh/max. EU data residency (Standard processing only). OpenAI's better default than Astra on cost. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "automatic", "reasoning_effort": ["none", "low", "medium", "high", "xhigh", "max"], "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/batch", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 5.0}, "priority": {"input_per_1m": 4.0, "output_per_1m": 20.0}}, "context_step": {"above_tokens": 272000, "input_per_1m": 4.0, "output_per_1m": 15.0}, "reasoning_output_per_1m": null, "effort_supported": ["none", "xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "gpt-6-luna", "provider": "openai", "display_name": "GPT-6 Luna", "status": "active", "released_at": "2026-09-14", "context_window": 1050000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.5, "cached_input_per_1m": 0.01, "cache_write_per_1m": 0.125}, "features": ["tool_use", "vision", "prompt_caching", "structured_output", "reasoning"], "tier": "micro", "capability_profile": ["tool_use", "long_context", "speed", "vision", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": ["ingestion", "triage", "high_volume", "summarisation", "long_context"], "bad_for": ["deep_reasoning", "complex_planning"], "latency": "low"}, "aedelgard_note": "GPT-6 efficient tier (2026-09-14): $0.10/$0.50 \u2014 HALVES GPT-5.6 Luna and is the cheapest 1M-context model in the catalogue. 1.05M/128k, cutoff May 18 2026, effort none->max. Excellent for ingestion, triage and summarisation at volume; do not route planning here. [read 2026-09-23]", "price_confirmed_at": "2026-09-23", "_auto_classified": false, "caching": "automatic", "reasoning_effort": ["none", "low", "medium", "high", "xhigh", "max"], "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/batch", "/v1/responses", "/v1/decisions"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.05, "output_per_1m": 0.25}, "priority": {"input_per_1m": 0.2, "output_per_1m": 1.0}}, "context_step": {"above_tokens": 272000, "input_per_1m": 0.2, "output_per_1m": 0.75}, "reasoning_output_per_1m": null, "effort_supported": ["none", "xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "anthropic.claude-opus-5-5", "provider": "bedrock", "display_name": "Claude Opus 5.5", "status": "active", "released_at": "2026-09-23", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Bedrock-hosted alias of claude-opus-5-5. Use when IAM auth is required. Pricing is Bedrock's own and is NOT confirmed here \u2014 route to the native provider for cost-critical work. [read 2026-09-23]", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-6-sol", "provider": "bedrock", "display_name": "GPT-6 Sol", "status": "active", "released_at": "2026-09-23", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Bedrock-hosted alias of gpt-6-sol. Use when IAM auth is required. Pricing is Bedrock's own and is NOT confirmed here \u2014 route to the native provider for cost-critical work. [read 2026-09-23]", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 4.4, "output_per_1m": 16.5}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-6-luna", "provider": "bedrock", "display_name": "GPT-6 Luna", "status": "active", "released_at": "2026-09-23", "context_window": 300000, "cost": {"input_per_1m": null, "output_per_1m": null, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "micro", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Bedrock-hosted alias of gpt-6-luna. Use when IAM auth is required. Pricing is Bedrock's own and is NOT confirmed here \u2014 route to the native provider for cost-critical work. [read 2026-09-23]", "price_confirmed_at": null, "_auto_classified": false, "caching": "none", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 0.22, "output_per_1m": 0.825}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-5.4", "provider": "bedrock", "display_name": "GPT 5.4", "status": "active", "released_at": "2026-09-24", "context_window": 300000, "cost": {"input_per_1m": 2.75, "output_per_1m": 16.5, "cached_input_per_1m": 0.275, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:bedrock_mantle/openai.gpt-5.4. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:bedrock_mantle/openai.gpt-5.4", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 5.5, "output_per_1m": 24.75}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-5.5", "provider": "bedrock", "display_name": "GPT 5.5", "status": "active", "released_at": "2026-09-24", "context_window": 300000, "cost": {"input_per_1m": 5.5, "output_per_1m": 33.0, "cached_input_per_1m": 0.55, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-25): tier=flagship from family/size; price from litellm:bedrock_mantle/openai.gpt-5.5. Editorial review welcome.", "price_confirmed_at": "2026-09-25", "_auto_classified": true, "price_source": "litellm:bedrock_mantle/openai.gpt-5.5", "price_fetched_at": "2026-09-25T15:44:23Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 11.0, "output_per_1m": 49.5}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "claude-sonnet-5-5", "provider": "anthropic", "display_name": "Claude Sonnet 5.5", "status": "active", "released_at": "2026-09-29", "context_window": 1000000, "cost": {"input_per_1m": 2.0, "output_per_1m": 10.0, "cached_input_per_1m": 0.1, "cache_write_per_1m": 2.5}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "flagship", "capability_profile": ["tool_use", "long_context", "structured_output"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-29): tier=flagship from family/size; price from litellm:claude-sonnet-5-5 [unscoped source \u2014 treat price as provisional]. Editorial review welcome.", "price_confirmed_at": "2026-10-11", "_auto_classified": false, "price_source": "litellm:claude-sonnet-5-5", "price_fetched_at": "2026-09-29T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 5.0}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "anthropic.claude-sonnet-5-5", "provider": "bedrock", "display_name": "Claude Sonnet 5.5", "status": "active", "released_at": "2026-09-29", "context_window": 300000, "cost": {"input_per_1m": 2.2, "output_per_1m": 11.0, "cached_input_per_1m": 0.22, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-29): tier=flagship from family/size; price from litellm:bedrock_mantle/anthropic.claude-sonnet-5-5. Editorial review welcome.", "price_confirmed_at": "2026-09-29", "_auto_classified": true, "price_source": "litellm:bedrock_mantle/anthropic.claude-sonnet-5-5", "price_fetched_at": "2026-09-29T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "xai.grok-4.7", "provider": "bedrock", "display_name": "Grok 4.7", "status": "active", "released_at": "2026-09-29", "context_window": 300000, "cost": {"input_per_1m": 2.2, "output_per_1m": 6.6, "cached_input_per_1m": 0.55, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-29): tier=flagship from family/size; price from aws-price-list:bedrock@eu-north-1. Editorial review welcome.", "price_confirmed_at": "2026-09-29", "_auto_classified": true, "price_source": "aws-price-list:bedrock@eu-north-1", "price_fetched_at": "2026-09-29T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 500000, "source": "models.dev"}, "purpose": "text"}, {"id": "gpt-6.1-sol", "provider": "openai", "display_name": "gpt-6.1-sol", "status": "active", "released_at": "2026-09-30", "context_window": 128000, "cost": {"input_per_1m": 2.0, "output_per_1m": 10.0, "cached_input_per_1m": 0.1, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-30): tier=flagship from family/size; price from litellm:gpt-6.1-sol [unscoped source \u2014 treat price as provisional]. Editorial review welcome.", "price_confirmed_at": "2026-09-30", "_auto_classified": true, "price_source": "litellm:gpt-6.1-sol", "price_fetched_at": "2026-09-30T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/batch", "/v1/responses"], "max_output_tokens": 128000, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 1.0, "output_per_1m": 5.0}, "priority": {"input_per_1m": 4.0, "output_per_1m": 20.0}}, "context_step": {"above_tokens": 272000, "input_per_1m": 4.0, "output_per_1m": 15.0}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "openai.gpt-6.1-sol", "provider": "bedrock", "display_name": "GPT-6.1 Sol", "status": "active", "released_at": "2026-09-30", "context_window": 300000, "cost": {"input_per_1m": 2.2, "output_per_1m": 11.0, "cached_input_per_1m": 0.11, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-09-30): tier=flagship from family/size; price from litellm:bedrock_mantle/openai.gpt-6.1-sol. Editorial review welcome.", "price_confirmed_at": "2026-09-30", "_auto_classified": true, "price_source": "litellm:bedrock_mantle/openai.gpt-6.1-sol", "price_fetched_at": "2026-09-30T06:00:48Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": ["/v1/chat/completions", "/v1/responses"], "max_output_tokens": 131072, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 272000, "input_per_1m": 4.4, "output_per_1m": 16.5}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "amazon.nova-2-5-sonic", "provider": "bedrock", "display_name": "Nova 2.5 Sonic", "status": "active", "released_at": "2026-10-06", "context_window": 300000, "cost": {"input_per_1m": 0.363, "output_per_1m": 2.992, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from aws-price-list:bedrock@eu-north-1. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "aws-price-list:bedrock@eu-north-1", "price_fetched_at": "2026-10-07T06:00:48Z", "purpose": "text"}, {"id": "zai.glm-5.3", "provider": "bedrock", "display_name": "GLM 5.3", "status": "active", "released_at": "2026-10-07", "context_window": 300000, "cost": {"input_per_1m": 1.2, "output_per_1m": 3.84, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=flagship from family/size; price from aws-price-list:bedrock@eu-north-1. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "aws-price-list:bedrock@eu-north-1", "price_fetched_at": "2026-10-07T06:00:48Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "codestral-2508", "provider": "mistral", "display_name": "codestral-2508", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 0.9, "cached_input_per_1m": 0.03, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use", "reasoning"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/codestral-2508. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/codestral-2508", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "codestral-latest", "provider": "mistral", "display_name": "codestral-2508", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 0.9, "cached_input_per_1m": 0.03, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use", "reasoning"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/codestral-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/codestral-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-code-latest", "provider": "mistral", "display_name": "codestral-2508", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 0.9, "cached_input_per_1m": 0.03, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-code-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-code-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-code-fim-latest", "provider": "mistral", "display_name": "codestral-2508", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.3, "output_per_1m": 0.9, "cached_input_per_1m": 0.03, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-code-fim-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-code-fim-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "ministral-14b-2512", "provider": "mistral", "display_name": "ministral-14b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.2, "output_per_1m": 0.2, "cached_input_per_1m": 0.02, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/ministral-14b-2512. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-14b-2512", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "ministral-14b-latest", "provider": "mistral", "display_name": "ministral-14b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.2, "output_per_1m": 0.2, "cached_input_per_1m": 0.02, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/ministral-14b-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-14b-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "ministral-3b-2512", "provider": "mistral", "display_name": "ministral-3b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.1, "cached_input_per_1m": 0.01, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/ministral-3b-2512. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-3b-2512", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "ministral-3b-latest", "provider": "mistral", "display_name": "ministral-3b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.1, "cached_input_per_1m": 0.01, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/ministral-3b-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-3b-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 131072, "source": "models.dev"}, "purpose": "text"}, {"id": "ministral-8b-2512", "provider": "mistral", "display_name": "ministral-8b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.15, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/ministral-8b-2512. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-8b-2512", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "ministral-8b-latest", "provider": "mistral", "display_name": "ministral-8b-2512", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.15, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/ministral-8b-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/ministral-8b-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium-latest", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium-3-5", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium-3-5. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium-3-5", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium-3.5", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium-3.5. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium-3.5", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium-3", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium-3. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium-3", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-medium-2604", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-medium-2604. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-medium-2604", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-vibe-cli-latest", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-vibe-cli-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-vibe-cli-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "mistral-vibe-cli-with-tools", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/mistral-vibe-cli-with-tools. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-vibe-cli-with-tools", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "magistral-medium-latest", "provider": "mistral", "display_name": "mistral-medium-latest", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": 0.15, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use", "reasoning"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:mistral/magistral-medium-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/magistral-medium-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-small-2603", "provider": "mistral", "display_name": "mistral-small-2603", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.6, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/mistral-small-2603. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-small-2603", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-small-latest", "provider": "mistral", "display_name": "mistral-small-2603", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.6, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/mistral-small-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-small-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "mistral-vibe-cli-fast", "provider": "mistral", "display_name": "mistral-small-2603", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.6, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/mistral-vibe-cli-fast. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/mistral-vibe-cli-fast", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "magistral-small-latest", "provider": "mistral", "display_name": "mistral-small-2603", "status": "active", "released_at": "2026-10-07", "context_window": 256000, "cost": {"input_per_1m": 0.15, "output_per_1m": 0.6, "cached_input_per_1m": 0.015, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use", "reasoning"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/magistral-small-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/magistral-small-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "purpose": "text"}, {"id": "voxtral-small-2507", "provider": "mistral", "display_name": "voxtral-small-2507", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.4, "cached_input_per_1m": 0.01, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/voxtral-small-2507. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/voxtral-small-2507", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "audio"], "output": ["text"], "endpoints": null, "max_output_tokens": 32768, "source": "models.dev"}, "purpose": "text"}, {"id": "voxtral-small-latest", "provider": "mistral", "display_name": "voxtral-small-2507", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.4, "cached_input_per_1m": 0.01, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:mistral/voxtral-small-latest. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:mistral/voxtral-small-latest", "price_fetched_at": "2026-10-07T06:56:28Z", "modality": {"input": ["text", "audio"], "output": ["text"], "endpoints": null, "max_output_tokens": 32768, "source": "models.dev"}, "purpose": "text"}, {"id": "Qwen/Qwen3.8-27B", "provider": "nebius", "display_name": "Qwen3.8 27B", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.45, "output_per_1m": 3.0, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "streaming", "long_context"], "tier": "fast-flagship", "capability_profile": ["tool_use", "long_context"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from litellm:nebius/Qwen/Qwen3.8-27B. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "price_source": "litellm:nebius/Qwen/Qwen3.8-27B", "price_fetched_at": "2026-10-07T08:57:56Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 262144, "source": "models.dev"}, "purpose": "text"}, {"id": "gemini-nano-banana-2.1", "provider": "google", "display_name": "Nano Banana 2.1", "status": "active", "released_at": "2026-10-07", "context_window": 65536, "cost": {"input_per_1m": 1.5, "output_per_1m": 7.5, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["tool_use", "vision", "long_context"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from litellm:gemini/gemini-nano-banana-2.1. Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": null, "price_source": "litellm:gemini/gemini-nano-banana-2.1", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["image", "text", "pdf"], "output": ["text", "image"], "endpoints": ["/v1/chat/completions", "/v1/images/generations", "/v1/images/edits"], "max_output_tokens": 32768, "source": "litellm+models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.75, "output_per_1m": 3.75}}, "context_step": null, "reasoning_output_per_1m": null, "effort_supported": null, "source": "litellm", "supports_reasoning": true}, "purpose": "image"}, {"id": "berget:Qwen/Qwen3.8-27B-FP8", "provider": "berget", "display_name": "Qwen3.8-27B-FP8", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.4471, "output_per_1m": 3.3531, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["streaming", "tool_use", "vision", "structured_output", "long_context"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from berget-api (EUR 0.4/3 per 1M, USD at ECB 1.1177 on 2026-10-07). Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": {"input_per_1m": 0.4482, "output_per_1m": 3.3618, "cached_input_per_1m": null, "cache_write_per_1m": null, "source": "berget-api (EUR 0.4/3 per 1M, USD at ECB 1.1206 on 2026-10-09)"}, "price_source": "berget-api (EUR 0.4/3 per 1M, USD at ECB 1.1177 on 2026-10-07)", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 65536, "source": "models.dev"}, "purpose": "text"}, {"id": "berget:mistralai/Mistral-Small-3.2-24B-Instruct-2506", "provider": "berget", "display_name": "Mistral-Small-3.2-24B-Instruct-2506", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.3353, "output_per_1m": 0.3353, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["streaming", "tool_use", "structured_output", "long_context"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from berget-api (EUR 0.3/0.3 per 1M, USD at ECB 1.1177 on 2026-10-07). Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": {"input_per_1m": 0.3362, "output_per_1m": 0.3362, "cached_input_per_1m": null, "cache_write_per_1m": null, "source": "berget-api (EUR 0.3/0.3 per 1M, USD at ECB 1.1206 on 2026-10-09)"}, "price_source": "berget-api (EUR 0.3/0.3 per 1M, USD at ECB 1.1177 on 2026-10-07)", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["text"], "output": ["text"], "endpoints": null, "max_output_tokens": 8192, "source": "models.dev"}, "purpose": "text"}, {"id": "berget:zai-org/GLM-5.3-Flash", "provider": "berget", "display_name": "GLM-5.3-Flash", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.2794, "output_per_1m": 0.5588, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["streaming", "tool_use", "vision", "structured_output", "long_context"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast from family/size; price from berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1177 on 2026-10-07). Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": {"input_per_1m": 0.2802, "output_per_1m": 0.5603, "cached_input_per_1m": null, "cache_write_per_1m": null, "source": "berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1206 on 2026-10-09)"}, "price_source": "berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1177 on 2026-10-07)", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["text", "image", "video"], "output": ["text"], "endpoints": null, "max_output_tokens": 16384, "source": "models.dev"}, "purpose": "text"}, {"id": "berget:moonshotai/Kimi-K3", "provider": "berget", "display_name": "Kimi-K3", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 3.3531, "output_per_1m": 16.7655, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["streaming", "tool_use", "vision", "structured_output", "long_context"], "tier": "flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=flagship from family/size; price from berget-api (EUR 3/15 per 1M, USD at ECB 1.1177 on 2026-10-07). Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": {"input_per_1m": 3.3618, "output_per_1m": 16.809, "cached_input_per_1m": null, "cache_write_per_1m": null, "source": "berget-api (EUR 3/15 per 1M, USD at ECB 1.1206 on 2026-10-09)"}, "price_source": "berget-api (EUR 3/15 per 1M, USD at ECB 1.1177 on 2026-10-07)", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["text", "image"], "output": ["text"], "endpoints": null, "max_output_tokens": 32768, "source": "models.dev"}, "purpose": "text"}, {"id": "berget:google/gemma-4-31B-it", "provider": "berget", "display_name": "gemma-4-31B-it", "status": "active", "released_at": "2026-10-07", "context_window": 128000, "cost": {"input_per_1m": 0.2794, "output_per_1m": 0.5588, "cached_input_per_1m": null, "cache_write_per_1m": null}, "features": ["streaming", "tool_use", "vision", "structured_output", "long_context"], "tier": "fast-flagship", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-07): tier=fast-flagship from family/size; price from berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1177 on 2026-10-07). Editorial review welcome.", "price_confirmed_at": "2026-10-07", "_auto_classified": true, "_provider_price": {"input_per_1m": 0.2802, "output_per_1m": 0.5603, "cached_input_per_1m": null, "cache_write_per_1m": null, "source": "berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1206 on 2026-10-09)"}, "price_source": "berget-api (EUR 0.25/0.5 per 1M, USD at ECB 1.1177 on 2026-10-07)", "price_fetched_at": "2026-10-07T16:57:07Z", "modality": {"input": ["audio", "image", "text", "video"], "output": ["text"], "endpoints": null, "max_output_tokens": 8192, "source": "models.dev"}, "purpose": "text"}, {"id": "claude-haiku-5-5", "provider": "anthropic", "display_name": "Claude Haiku 5.5", "status": "active", "released_at": "2026-10-08", "context_window": 1000000, "cost": {"input_per_1m": 0.1, "output_per_1m": 0.5, "cached_input_per_1m": 0.01, "cache_write_per_1m": 0.125}, "features": ["tool_use", "prompt_caching", "vision"], "tier": "fast", "capability_profile": ["tool_use", "speed"], "cost_efficiency": "high", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-08): tier=fast from family/size; price from litellm:claude-haiku-5-5 [unscoped source \u2014 treat price as provisional]. Editorial review welcome.", "price_confirmed_at": "2026-10-11", "_auto_classified": false, "_provider_price": null, "price_source": "litellm:claude-haiku-5-5", "price_fetched_at": "2026-10-08T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": {"batch": {"input_per_1m": 0.05, "output_per_1m": 0.25}}, "context_step": {"above_tokens": 100000, "input_per_1m": 0.5, "output_per_1m": 2.5}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}, {"id": "anthropic.claude-haiku-5-5", "provider": "bedrock", "display_name": "Claude Haiku 5.5", "status": "active", "released_at": "2026-10-08", "context_window": 300000, "cost": {"input_per_1m": 0.11, "output_per_1m": 0.55, "cached_input_per_1m": 0.011, "cache_write_per_1m": null}, "features": ["tool_use"], "tier": "fast", "capability_profile": [], "cost_efficiency": "medium", "classification": {"good_for": [], "bad_for": [], "latency": "unknown"}, "aedelgard_note": "Auto-classified (2026-10-08): tier=fast from family/size; price from litellm:bedrock_mantle/anthropic.claude-haiku-5-5. Editorial review welcome.", "price_confirmed_at": "2026-10-08", "_auto_classified": true, "_provider_price": null, "price_source": "litellm:bedrock_mantle/anthropic.claude-haiku-5-5", "price_fetched_at": "2026-10-08T06:00:48Z", "modality": {"input": ["text", "image", "pdf"], "output": ["text"], "endpoints": null, "max_output_tokens": 128000, "source": "models.dev"}, "cost_ladder": {"service_tiers": null, "context_step": {"above_tokens": 100000, "input_per_1m": 0.55, "output_per_1m": 2.75}, "reasoning_output_per_1m": null, "effort_supported": ["xhigh", "max"], "source": "litellm", "supports_reasoning": true}, "purpose": "text"}], "count": 117, "updated_at": "2026-10-12T06:00:48Z", "ttl_seconds": 3600}