[{"api":"Z.ai API; self-hostable via vLLM/SGLang/Transformers","benchmarks":"[{\"kind\":\"Model-card reported\",\"name\":\"Terminal-Bench 2.1\",\"score\":\"84.3%\",\"setting\":\"Temperature 1; max 65,536; 6-hour limit\"},{\"kind\":\"Model-card reported\",\"name\":\"DeepSWE\",\"score\":\"63.4%\",\"setting\":\"Temperature .95; max 400,000; 6-hour limit\"},{\"kind\":\"Model-card reported\",\"name\":\"Humanity's Last Exam\",\"score\":\"55.3%\",\"setting\":\"Temperature 1; max 163,840; 300k context\"}]","context":"300,000 tokens in retained evaluation setting","id":"zai:glm-5.3-flash","input_price":"Not published in retained source","latency":"Not published in retained source","license":"MIT","model":"GLM-5.3-Flash","openness":"Open-weight and model materials under MIT","output_price":"Not published in retained source","parameters":"320B total / 18B active","provider":"Z.ai","release_date":"2026-09-19","source":"https://huggingface.co/zai-org/GLM-5.3-Flash","source_ids":"[\"glm53-card\"]","throughput":"Not published in retained source","version":"GLM-5.3-Flash"},{"api":"Gemini API","benchmarks":"[]","context":"1,048,576 input / 65,536 output tokens","id":"google:gemini-3.8-flash","input_price":"$0.75 / 1M tokens through 2026-12-31; $1.50 from 2027-01-01","latency":"Not published in retained source","license":"Google API terms","model":"Gemini 3.8 Flash","openness":"Proprietary API model","output_price":"$3.75 / 1M tokens through 2026-12-31; $7.50 from 2027-01-01","parameters":"Not published in retained source","provider":"Google","release_date":"2026-09-17 page update; release date not stated","source":"https://ai.google.dev/gemini-api/docs/models/gemini-3.8-flash","source_ids":"[\"google-gemini38\",\"google-gemini38-pricing\"]","throughput":"Not published in retained source","version":"gemini-3.8-flash"},{"api":"DeepSeek API (`deepseek-flash`); provider links weights","benchmarks":"[]","context":"Not published in retained source","id":"deepseek:deepseek-v4.1-flash","input_price":"Not published in retained source","latency":"Not published in retained source","license":"Not published in retained source","model":"DeepSeek V4.1-Flash","openness":"Open-weight status indicated by provider; license not verified in retained source","output_price":"Not published in retained source","parameters":"552B total / 8B input-active / 16B output-active","provider":"DeepSeek","release_date":"2026-09-10","source":"https://www.deepseek.com/en/news/deepseek-v4-1-flash/","source_ids":"[\"deepseek-v41\"]","throughput":"Not published in retained source","version":"deepseek-flash"},{"api":"OpenAI API; Azure; Amazon Bedrock","benchmarks":"[{\"kind\":\"Provider reported\",\"name\":\"Terminal-Bench 4.0\",\"score\":\"57.9%\",\"setting\":\"OpenAI evaluation; provider setting retained in launch report\"},{\"kind\":\"Provider reported\",\"name\":\"GPQA Diamond\",\"score\":\"96.0%\",\"setting\":\"OpenAI evaluation; provider setting retained in launch report\"}]","context":"1,050,000 input / 131,072 output tokens","id":"openai:gpt-6-astra","input_price":"$10.00 / 1M tokens","latency":"Not published in retained source","license":"Provider terms","model":"GPT-6 Astra","openness":"Proprietary API model","output_price":"$50.00 / 1M tokens","parameters":"Not published in retained source","provider":"OpenAI","release_date":"2026-09-03","source":"https://developers.openai.com/api/docs/models/gpt-6-astra","source_ids":"[\"openai-api\",\"openai-gpt6-launch\"]","throughput":"Not published in retained source","version":"gpt-6-astra"},{"api":"Claude API; Amazon Web Services; Google Cloud; Microsoft Azure","benchmarks":"[{\"kind\":\"Provider reported\",\"name\":\"Terminal-Bench-Science 0.1\",\"score\":\"52.6%\",\"setting\":\"Production safeguards enabled; Anthropic setup\"},{\"kind\":\"Provider reported\",\"name\":\"Terminal-Bench 4.0\",\"score\":\"55.8%\",\"setting\":\"Production safeguards enabled; Anthropic setup\"},{\"kind\":\"Provider reported\",\"name\":\"Humanity's Last Exam (no tools)\",\"score\":\"60.9%\",\"setting\":\"Production safeguards enabled; no tools\"},{\"kind\":\"Provider reported\",\"name\":\"Humanity's Last Exam (with tools)\",\"score\":\"65.0%\",\"setting\":\"Production safeguards enabled; with tools\"},{\"kind\":\"Provider reported\",\"name\":\"CursorBench 3.2.0\",\"score\":\"73.4%\",\"setting\":\"Max effort; Anthropic report\"},{\"kind\":\"Provider reported\",\"name\":\"OSWorld 2.0 (partial)\",\"score\":\"77.9%\",\"setting\":\"August 2026 task release; partial completion; production safeguards enabled\"},{\"kind\":\"Provider reported\",\"name\":\"OSWorld 2.0 (strict)\",\"score\":\"41.7%\",\"setting\":\"August 2026 task release; strict completion; production safeguards enabled\"}]","context":"Not published in retained source","id":"anthropic:claude-fable-5.1","input_price":"$10.00 / 1M tokens","latency":"Not published in retained source","license":"Provider terms","model":"Claude Fable 5.1","openness":"Proprietary API model","output_price":"$50.00 / 1M tokens","parameters":"Not published in retained source","provider":"Anthropic","release_date":"2026-09-01","source":"https://www.anthropic.com/claude-fable-and-mythos-5-1","source_ids":"[\"anthropic-fable51\"]","throughput":"Not published in retained source","version":"claude-fable-5-1"},{"api":"Qwen Cloud API; self-hostable","benchmarks":"[{\"kind\":\"Model-card reported\",\"name\":\"Terminal-Bench 2.1\",\"score\":\"86.6%\",\"setting\":\"Qwen evaluation; harness and run settings retained in model card\"},{\"kind\":\"Model-card reported\",\"name\":\"DeepSWE\",\"score\":\"56.6%\",\"setting\":\"Qwen evaluation; harness and run settings retained in model card\"},{\"kind\":\"Model-card reported\",\"name\":\"SWE-bench Pro\",\"score\":\"67.7%\",\"setting\":\"Qwen evaluation; harness and run settings retained in model card\"},{\"kind\":\"Model-card reported\",\"name\":\"PaperBench\",\"score\":\"93.0%\",\"setting\":\"Qwen evaluation; harness and run settings retained in model card\"}]","context":"262,144 native / 1,010,000 extensible tokens","id":"qwen:qwen3.8-2.4t-a95b","input_price":"Not published in retained source","latency":"Not published in retained source","license":"Qwen3.8 Max custom license","model":"Qwen3.8 2.4T A95B","openness":"Open-weight; license is not OSI-identified in retained source","output_price":"Not published in retained source","parameters":"2.4T total / 95B active","provider":"Qwen","release_date":"2026-08-12 page update; release date not stated","source":"https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B","source_ids":"[\"qwen38-card\"]","throughput":"Not published in retained source","version":"Qwen3.8-2.4T-A95B"},{"api":"Mistral API; Amazon Bedrock; Microsoft Azure","benchmarks":"[]","context":"256,000 tokens","id":"mistral:mistral-large-3","input_price":"$0.50 / 1M tokens","latency":"Not published in retained source","license":"Apache 2.0","model":"Mistral Large 3","openness":"Open-weight","output_price":"$1.50 / 1M tokens","parameters":"675B total / 41B active","provider":"Mistral AI","release_date":"2025-12-02","source":"https://mistral.ai/news/mistral-3/","source_ids":"[\"mistral3\",\"mistral-pricing\"]","throughput":"Not published in retained source","version":"mistral-large-3"}]