Use case provisional data

Best AI models for agents: benchmark comparison

Using tools, following policies and finishing multi-step tasks. τ-bench simulates an airline support desk with rules and systems.

Direct answer: On τ-bench Verified (airline), the top score among the models on this page is Claude Fable 5, with 82%, per OpenRouter Benchmarks.

How to read this table

  1. Each row is a model. ClickTap the name to open the full page, with sources and version history. ClickTap a header to sort.
  2. Benchmark scores are percentages, and higher is better. I always compare the same metric, from the same source, in the latest version we checked. In this section, in the 90% to 100% range, I treat a gap of 1 or 2 points as possible measurement noise.
  3. The "?" next to each column explains what it measures and when it matters. "No data" means there is no measurement in the open sources this section uses. "Provisional data" marks a field derived by a rule, such as use case.
  4. Input and output prices are the OpenRouter list price, in US dollars per 1 million tokens, and the two can differ a lot. The "to check" badge marks an odd value in the source, which is left out of recommendations. Usage is the token share in OpenRouter's public ranking, for the latest day published by that source. Downloads are the official Hugging Face repository's 30-day window. Epoch's GPQA Diamond and OpenRouter's GPQA Diamond measure the same test, but they can differ.
Agents ClickTap a header to sort.
?what it is?????????
Claude Fable 5AnthropicclosedUS$ 10.00US$ 50.001,000,00085.9%no data82%70.7%no datano data
DeepSeek V4 Flash 0423DeepSeekopenUS$ 0.09US$ 0.181,048,576no datano data81.7%no data2.85%1,168,829
Gemini 3.7 FlashGoogleclosedUS$ 0.75US$ 3.751,048,57694.8%no data80.7%69.2%no datano data
Qwen3.5 397B A17BAlibabaopenUS$ 0.60US$ 3.60262,14486.4%no data80.2%no datano data826,220
GPT-5.1OpenAIclosedUS$ 1.25US$ 10.00400,00087.6%68%80%48%no datano data
Ling 3.1 FlashAnt GroupundisclosedUS$ 0.045US$ 0.141,000,000no datano data79.3%no datano datano data
Claude Fable 5.1AnthropicclosedUS$ 10.00US$ 50.001,000,000no datano data79.2%70.8%no datano data
Qwen3.6 27BAlibabaopenUS$ 0.32US$ 3.20262,14485.9%no data79%no datano data2,365,084
MiMo V2.6 ProXiaomiopenUS$ 0.43US$ 0.871,050,000no datano data78.9%no data0.54%no data
GLM-5.3Zhipu AI (Z.ai)openUS$ 0.07US$ 1.401,048,57690.9%no data78.8%41%1.55%no data
Claude Sonnet 5.5AnthropicclosedUS$ 2.00US$ 10.001,000,00095.6%no data78.8%46.5%1.18%no data
Kimi K3.1 ProMoonshot AIopenUS$ 2.00US$ 10.001,048,57694.2%no data78.7%62.2%2.51%no data
Gemini 4 ArgonGoogleclosedUS$ 2.00US$ 10.001,048,57696.4%no data78.7%76%0.54%no data
Claude Opus 5AnthropicclosedUS$ 5.00US$ 25.001,000,00093.9%no data78.7%59.9%no datano data
Qwen3.5-35B-A3BAlibabaopenUS$ 0.14US$ 1.00262,14483.5%no data78.7%no datano data1,582,689
DeepSeek V4 ProDeepSeekopenUS$ 0.96US$ 1.911,048,57691.7%77.6%78.6%52.9%0.66%no data
Gemini 3.5 Flash LiteGoogleclosedUS$ 0.30US$ 2.501,048,57683.3%no data78.3%no datano datano data
MiMo-V2.6-FlashXiaomiopenUS$ 0.14US$ 0.281,050,000no datano data78.2%no data2.22%76,859
Qwen3.8 27BAlibabaopenUS$ 0.15US$ 1.881,048,575no datano data78%no data0.34%7,443,748
DeepSeek V4.1 FlashDeepSeekopenUS$ 0.022US$ 0.401,048,57693.2%no data77.8%42.9%32.17%no data
Claude Opus 4.7AnthropicclosedUS$ 5.00US$ 25.001,000,00090.2%83.5%77.8%51.7%no datano data
Qwen3.5-122B-A10BAlibabaopenUS$ 0.26US$ 2.08262,144no datano data77.7%no datano data880,973
Hy4TencentopenUS$ 0.42US$ 1.251,048,576no datano data77.6%no data10.74%no data
Claude Opus 4.8AnthropicclosedUS$ 5.00US$ 25.001,000,00091%no data77.5%53%no datano data
Qwen3.8 2.4T A95BAlibabaopenUS$ 2.00US$ 6.001,048,576no datano data77.4%no datano data26,226
Kimi K2.7 CodeMoonshot AIopenUS$ 0.60US$ 3.50262,14487.9%no data77.3%36.5%no data91,687
Muse Glimmer 30BMetaopenUS$ 0.30US$ 1.20131,072no datano data77.2%no datano datano data
Gemma 4 31BGoogleopenUS$ 0.20US$ 0.40262,14475.8%no data77%10.4%0.22%no data
Gemini 3.1 ProGoogleclosedUS$ 2.00US$ 12.001,048,57694.4%75.6%77%73.5%no datano data
GLM 5Zhipu AI (Z.ai)openUS$ 1.00US$ 3.20204,80087.8%72.1%76.9%no datano data1,370,207
Gemini 3.8 FlashGoogleclosedUS$ 0.82US$ 4.121,048,57695.4%no data76.7%69.7%0.82%no data
GLM-5.2Zhipu AI (Z.ai)openUS$ 0.007 to checkUS$ 6.801,048,57691.9%78.7%76.7%34.2%0.43%no data
Nova Micro 1.0AmazonundisclosedUS$ 0.035US$ 0.14128,000no datano data76.7%no datano datano data
DeepSeek V4 Flash Vision ExpDeepSeekopenUS$ 0.44US$ 1.321,048,576no datano data76.7%no datano data148,215
Claude Opus 4.5AnthropicclosedUS$ 5.00US$ 25.00200,00086%76.7%76.6%45.7%no datano data
Qwen3.8 FlashAlibabaopenUS$ 0.15US$ 0.471,000,000no datano data76.4%no data0.7%no data
GLM-5.3-FlashZhipu AI (Z.ai)openUS$ 0.04US$ 0.0781,048,57690.2%no data76.3%no data5.09%no data
MiniMax M3MiniMaxopenUS$ 0.30US$ 1.201,048,57690.9%no data76.3%no data0.3%no data
Gemini 3.6 FlashGoogleclosedUS$ 0.75US$ 3.751,048,57694.1%no data76.3%66.2%no datano data
Kimi K3Moonshot AIopenUS$ 0.19 to checkUS$ 16.001,048,57693.1%no data76.2%50.6%0.5%no data
Qwen3.6 35B A3BAlibabaopenUS$ 0.10US$ 0.90262,14484.8%no data76.2%no datano data4,177,044
GLM 4.7Zhipu AI (Z.ai)openUS$ 0.60US$ 2.20204,80083.3%no data76%32.2%no data47,416
Claude Opus 4.6AnthropicclosedUS$ 5.00US$ 25.001,000,00090.5%78.7%75.8%47%no datano data
GLM 5.1Zhipu AI (Z.ai)openUS$ 1.40US$ 4.40204,80089.9%74.2%75.8%34%no data867,145
Muse Spark OpenMetaopenUS$ 1.25US$ 4.251,048,57688.6%no data75.7%56.6%no datano data
Qwen3.5-9BAlibabaopenUS$ 0.052US$ 0.077262,14479%no data75.7%no datano data6,370,574
Claude Sonnet 5AnthropicclosedUS$ 2.00US$ 10.001,000,00090.5%no data75.4%33.7%0.25%no data
Claude Opus 5.5AnthropicclosedUS$ 4.00US$ 20.001,000,00090.6%no data75.3%72.2%3.21%no data
Grok 4.7xAIclosedUS$ 2.00US$ 6.00500,00092.7%no data75.3%56%0.22%no data
Grok 4.6xAIclosedUS$ 2.00US$ 6.00500,00094%no data75.3%49.3%no datano data
GPT-5.3-CodexOpenAIclosedUS$ 1.75US$ 14.00400,000no data74.8%75.3%no datano datano data
GPT-5.4OpenAIclosedUS$ 2.50US$ 15.001,050,00093.3%76.9%75%45.1%no datano data
Nemotron 3.5 Super VLNVIDIAopenUS$ 0.15US$ 0.471,048,576no datano data74.9%no data1.49%no data
Claude Sonnet 4.5AnthropicclosedUS$ 3.00US$ 15.001,000,00082.3%71.3%74.9%30.7%no datano data
Kimi K2.6Moonshot AIopenUS$ 0.65US$ 3.41262,14490.8%76.7%74.9%34.9%no datano data
Claude Sonnet 4.6AnthropicclosedUS$ 3.00US$ 15.001,000,00087.4%75.2%74.8%35.5%no datano data
MiniMax M3.1 FlashMiniMaxopenUS$ 0.15US$ 0.601,048,576no datano data74.7%no data1.89%55,087
GPT-5.5OpenAIclosedUS$ 5.00US$ 30.001,050,00094%80.6%74.7%63%no datano data
DeepSeek V3.2DeepSeekopenUS$ 0.26US$ 0.38163,84083.4%no data74.7%no datano datano data
MiMo-V2.5XiaomiopenUS$ 0.40US$ 2.001,000,000no datano data74.7%no datano data100,669
GPT-5.6 SolOpenAIclosedUS$ 4.40US$ 22.001,050,00093.5%no data74.2%69.7%no datano data
MiMo-V2.5-ProXiaomiopenUS$ 0.48US$ 1.801,024,000no datano data74.1%no datano data9,642
Claude Haiku 5.5AnthropicclosedUS$ 0.10 to checkUS$ 0.501,000,00089.6%no data74%23.8%1.27%no data
Ling 3.0 FlashAnt GroupopenUS$ 0.021US$ 0.063262,144no datano data74%no datano data9,515
Kimi K2.5Moonshot AIopenUS$ 0.53US$ 3.33262,14487.6%73.8%73.8%34.3%no data393,277
MiniMax M2.7MiniMaxopenUS$ 0.21US$ 0.84196,608no datano data73.4%no datano data303,481
Gemini 3 Flash PreviewGoogleundisclosedUS$ 0.50US$ 3.001,048,576no datano data73.3%no data0.32%no data
GLM 4.6Zhipu AI (Z.ai)openUS$ 0.50US$ 2.00204,800no datano data73.3%no datano data27,796
DeepSeek V3.1DeepSeekopenUS$ 0.25US$ 0.95163,840no datano data73.1%no datano data266,086
DeepSeek V3.1 TerminusDeepSeekopenUS$ 0.27US$ 1.00163,840no datano data73%no datano data9,250
Gemini 3.5 FlashGoogleclosedUS$ 1.50US$ 9.001,048,57692.8%79.3%73%66.2%no datano data
GPT-5.6 LunaOpenAIclosedUS$ 0.20US$ 1.201,050,00091.6%no data72.7%41%0.68%no data
Gemini 3.1 Flash LiteGoogleclosedUS$ 0.25US$ 1.501,048,57681.8%no data72.7%no data0.26%no data
GPT-6 Astra ProOpenAIundisclosedUS$ 10.00US$ 50.001,050,000no datano data72.7%no datano datano data
Qwen3.7 PlusAlibabaclosedUS$ 0.32US$ 1.281,000,00087.9%no data72.7%no datano datano data
GPT-5 MiniOpenAIclosedUS$ 0.25US$ 2.00400,00075%64.7%72.4%21.6%no datano data
GPT-5.6 TerraOpenAIclosedUS$ 2.00US$ 12.001,050,00093.3%no data72.3%43.2%no datano data
Ling 3.0 Flash VLAnt GroupopenUS$ 0.06US$ 0.18131,072no datano data72%no datano data9,945
Qwen3.7 MaxAlibabaclosedUS$ 1.48US$ 4.421,000,00090.9%77.3%72%55.8%no datano data
GPT-5.6 Sol ProOpenAIundisclosedUS$ 4.00US$ 20.001,050,000no datano data71.7%no datano datano data
Nemotron 3.5 LightningNVIDIAopenUS$ 0.07US$ 0.20262,144no datano data71.5%no data0.52%1,070,415
GPT-6.1 SolOpenAIclosedUS$ 2.00US$ 10.001,050,00095.4%no data70.7%73.9%1.62%no data
GPT-5OpenAIclosedUS$ 1.25US$ 10.00400,00086.2%73.6%70.7%50.1%no datano data
GPT-5.6 Luna ProOpenAIundisclosedUS$ 0.20US$ 1.201,050,000no datano data70.7%no datano datano data
Gemma 4 26B A4BGoogleopenUS$ 0.08US$ 0.10262,14473.2%no data70.6%no data0.21%5,439,703
GPT-6 SolOpenAIclosedUS$ 2.00US$ 10.001,050,00094.3%no data70.2%60.7%no datano data
GPT-6.1 TerraOpenAIclosedUS$ 1.00US$ 5.001,050,00094.3%no data70%57.3%0.58%no data
GPT-6 AstraOpenAIclosedUS$ 10.00US$ 50.001,050,00095.8%no data69.7%75.6%no datano data
GLM 4.7 FlashZhipu AI (Z.ai)openUS$ 0.06US$ 0.40200,000no datano data69.7%no datano data1,071,754
GPT-6 Luna ProOpenAIundisclosedUS$ 0.10US$ 0.501,050,000no datano data69.3%no data0.31%no data
MiniMax M2.5MiniMaxopenUS$ 0.27US$ 1.08200,000no datano data69.2%no datano data1,184,681
GPT-6 LunaOpenAIclosedUS$ 0.10US$ 0.501,050,00090.5%no data68.4%41.4%3.42%no data
gpt-oss-120bOpenAIopenUS$ 0.15US$ 0.60131,07275.8%no data67.3%no data0.24%no data
GPT-5.4 MiniOpenAIclosedUS$ 0.75US$ 4.50400,00086.9%no data66.7%29.4%no datano data
Claude Haiku 4.5AnthropicclosedUS$ 1.00US$ 5.00200,00071.2%no data66.4%13.2%no datano data
GPT-5.4 NanoOpenAIclosedUS$ 0.20US$ 1.25400,00078.5%no data65.6%11.7%no datano data
Granite 4.2 8BIBMopenUS$ 0.06US$ 0.15131,072no datano data64%no datano datano data
Qwen3 235B A22B Thinking 2507AlibabaopenUS$ 0.45US$ 3.50128,000no datano data61.7%no datano data8,177
Qwen3 Coder NextAlibabaopenUS$ 0.12US$ 0.80262,144no datano data60.7%no datano data425,899
GLM 4.6VZhipu AI (Z.ai)openUS$ 0.30US$ 0.90131,072no datano data60.7%no datano data5,245
Gemini 2.5 ProGoogleclosedUS$ 1.25US$ 10.001,048,57685.3%57.6%60%no datano datano data
Gemini 2.5 FlashGoogleclosedUS$ 0.30US$ 2.501,048,576no datano data59.1%no datano datano data
DeepSeek R1DeepSeekopenUS$ 0.50US$ 2.15163,84076.3%no data58%no datano datano data
Qwen3 32BAlibabaopenUS$ 0.08US$ 0.28131,07265.7%no data54.3%no datano data3,001,273
Gemini 2.5 Flash LiteGoogleundisclosedUS$ 0.10US$ 0.401,048,576no datano data53.7%no data0.17%no data
gpt-oss-20bOpenAIopenUS$ 0.03US$ 0.13131,07260.8%no data53.5%no datano datano data
GPT-4.1OpenAIclosedUS$ 2.00US$ 8.001,047,57666.9%48.5%52.3%31.1%no datano data
Nemotron 3 Nano 30B A3BNVIDIAopenUS$ 0.06US$ 0.24262,144no datano data52%no datano data1,556,699
GPT-5 NanoOpenAIclosedUS$ 0.05US$ 0.40400,00069.4%no data48.3%11.7%no datano data
Qwen3-CoderAlibabaopenUS$ 0.30US$ 1.00262,144no datano data48%no datano data58,791
Gemma 3 12BGoogleopenUS$ 0.05US$ 0.15131,07239.5%no data48%no datano datano data
Qwen3 VL 235B A22B InstructAlibabaopenUS$ 0.21US$ 1.90262,144no datano data48%no datano data325,913
Qwen3-235B-A22BAlibabaopenUS$ 0.14US$ 0.80262,14480.1%no data47.9%40.4%no datano data
Qwen3 Next 80B A3B InstructAlibabaopenUS$ 0.09US$ 1.10262,144no datano data47.5%no datano data129,851
Qwen3 30B A3BAlibabaopenUS$ 0.12US$ 0.5040,96070.1%no data47%no datano data1,135,784
Mistral Small 4Mistral AIopenUS$ 0.15US$ 0.60262,144no datano data46%no datano data28,703
Qwen3 Coder 30B A3B InstructAlibabaopenUS$ 0.15US$ 0.60262,144no datano data46%no datano data1,679,625
GPT-4.1 MiniOpenAIclosedUS$ 0.40US$ 1.601,047,57665.8%no data45.1%12.7%no datano data
Qwen3 14BAlibabaopenUS$ 0.10US$ 0.2240,96063.8%no data44%no datano data12,907,991
DeepSeek V3DeepSeekopenUS$ 0.29US$ 1.14163,84067.6%no data43.8%no datano datano data
Llama 4 MaverickMetaopenUS$ 0.35US$ 1.00524,28867%no data43.4%no datano data11,953
GPT-4oOpenAIclosedUS$ 2.50US$ 10.00128,00049.2%31%43.2%26%no datano data
DeepSeek V3DeepSeekopenUS$ 0.32US$ 0.89163,84056.5%no data43%no datano data1,486,147
Qwen3 30B A3B Instruct 2507AlibabaopenUS$ 0.10US$ 0.30262,144no datano data41.5%no datano data880,864
Qwen3 VL 30B A3B InstructAlibabaopenUS$ 0.15US$ 0.60262,144no datano data41.3%no datano data446,889
Llama 3.3 70B InstructMetaopenUS$ 0.71US$ 0.71131,072no datano data39.9%no datano data349,408
Qwen3 VL 8B InstructAlibabaopenUS$ 0.25US$ 0.75262,144no datano data32.7%no datano data7,052,170
GPT-4o-miniOpenAIclosedUS$ 0.15US$ 0.60128,00037.7%no data30.7%8.3%no datano data
Qwen2.5 72B InstructAlibabaopenUS$ 0.36US$ 0.4032,768no datano data27.3%no datano data151,833
Llama 3.1 8B InstructMetaopenUS$ 0.02US$ 0.04131,072no datano data23.5%no datano datano data
GPT-4.1 NanoOpenAIclosedUS$ 0.10US$ 0.401,047,57648.9%no data14%6%no datano data
Step 5 PreviewStepFunopenUS$ 1.00US$ 2.701,000,000no datano datano datano data3.25%no data
Nemotron 3 UltraNVIDIAopenUS$ 0.60US$ 2.40262,14485.4%no datano datano data1.77%1,080,025
Muse Spark 1.3 ContributorMetaundisclosedUS$ 0.10US$ 0.201,048,576no datano datano datano data1.6%no data
Hy3TencentopenUS$ 0.13US$ 0.53262,144no datano datano datano data1.09%1,141,189
Solar Mini 4UpstageundisclosedUS$ 0.05US$ 0.20524,288no datano datano datano data0.59%no data
Qwen3.7 FlashAlibabaclosedUS$ 0.03US$ 0.131,000,00082.3%no datano datano data0.2%no data
Solar Pro 4UpstageundisclosedUS$ 0.15US$ 0.60524,288no datano datano datano data0.2%no data
Claude Fable 5.2AnthropicclosedUS$ 8.00US$ 40.001,000,00093.9%no datano data94%no datano data
GPT-5.2OpenAIclosedUS$ 1.75US$ 14.00400,00091.4%73.8%no data37.1%no datano data
Qwen3.8 MaxAlibabaclosedUS$ 2.00US$ 6.001,000,00092.7%no datano data47.3%no datano data
Mistral Large 4Mistral AIopenUS$ 0.70US$ 2.101,048,57690.2%no datano data22.4%no datano data
Nova 2 LiteAmazonundisclosedUS$ 0.30US$ 2.501,000,000no datano datano datano datano datano data
Nova Lite 1.0AmazonundisclosedUS$ 0.06US$ 0.24300,000no datano datano datano datano datano data
Nova Premier 1.0AmazonundisclosedUS$ 2.50US$ 12.501,000,000no datano datano datano datano datano data
Nova Pro 1.0AmazonundisclosedUS$ 0.80US$ 3.20300,000no datano datano datano datano datano data
Claude Opus 4.1AnthropicclosedUS$ 15.00US$ 75.00200,00077.3%73.3%no datano datano datano data
Seed 1.6ByteDanceundisclosedno datano datano datano datano datano datano datano datano data
Seed 1.6 FlashByteDanceundisclosedno datano datano datano datano datano datano datano datano data
Seed 2.1 TurboByteDanceundisclosedUS$ 0.50US$ 2.50262,144no datano datano datano datano datano data
Seed-2.0-CodeByteDanceundisclosedno datano datano datano datano datano datano datano datano data
Seed-2.0-LiteByteDanceundisclosedUS$ 0.25US$ 2.00262,144no datano datano datano datano datano data
Seed-2.0-MiniByteDanceundisclosedUS$ 0.10US$ 0.40262,144no datano datano datano datano datano data
Command ACohereopenUS$ 2.50US$ 10.00256,000no datano datano datano datano datano data
Command A+CohereundisclosedUS$ 0.30US$ 1.50192,000no datano datano datano datano datano data
Command R (08-2024)CohereundisclosedUS$ 0.15US$ 0.60128,000no datano datano datano datano datano data
Command R+ (08-2024)CohereundisclosedUS$ 2.50US$ 10.00128,000no datano datano datano datano datano data
Command R7B (12-2024)CohereundisclosedUS$ 0.037US$ 0.15128,000no datano datano datano datano datano data
DeepSeek V3.2 ExpDeepSeekopenUS$ 0.27US$ 0.41163,840no datano datano datano datano data45,262
Gemini 2.5 Pro Preview 06-05GoogleundisclosedUS$ 1.25US$ 10.001,048,576no datano datano datano datano datano data
Nano Banana Pro (Gemini 3 Pro Image)GoogleundisclosedUS$ 2.00US$ 12.00131,072no datano datano datano datano datano data
Gemini 3.1 Flash Lite PreviewGoogleundisclosedUS$ 0.25US$ 1.501,048,576no datano datano datano datano datano data
Gemini 3.1 Pro Preview Custom ToolsGoogleundisclosedUS$ 2.00US$ 12.001,048,576no datano datano datano datano datano data
Nano Banana 2.1GoogleundisclosedUS$ 1.50US$ 7.5065,536no datano datano datano datano datano data
Gemma 3 27BGoogleopenUS$ 0.08US$ 0.35131,07247.7%no datano datano datano datano data
Ling 3.0 TinyAnt GroupopenUS$ 0.015US$ 0.045262,144no datano datano datano datano data13,841
LFM2.5-2.6BLiquid AIopenUS$ 0.35US$ 0.50128,000no datano datano datano datano data47,737
LFM2.5-24B-A2BLiquid AIopenUS$ 0.05US$ 0.70128,000no datano datano datano datano data39,597
LFM3-1.7BLiquid AIopenUS$ 0.08US$ 0.20262,144no datano datano datano datano data82,718
LongCat 2.0MeituanopenUS$ 0.30US$ 1.201,048,756no datano datano datano datano data3,024
Llama 4 ScoutMetaopenUS$ 0.10US$ 0.301,310,72051.8%no datano datano datano data111,251
Muse EmberMetaundisclosedUS$ 0.10US$ 0.501,048,576no datano datano datano datano datano data
Muse Spark 1.1MetaclosedUS$ 1.25US$ 4.251,048,576no datano datano data57.8%no datano data
Muse Spark 1.2MetaclosedUS$ 1.25US$ 4.251,048,576no datano datano data60.3%no datano data
Muse Spark 1.2 ContributorMetaundisclosedUS$ 0.10US$ 0.201,048,576no datano datano datano datano datano data
Muse Spark 1.3MetaclosedUS$ 1.25US$ 4.251,048,576no datano datano datano datano datano data
Muse WonderMetaclosedUS$ 4.00US$ 20.001,048,576no datano datano datano datano datano data
Codestral 2508Mistral AIundisclosedUS$ 0.30US$ 0.90256,000no datano datano datano datano datano data
Devstral 2 2512Mistral AIopenUS$ 0.40US$ 2.00262,144no datano datano datano datano data6,656
Ministral 3 14B 2512Mistral AIopenUS$ 0.20US$ 0.20262,144no datano datano datano datano data180,348
Ministral 3 3B 2512Mistral AIopenUS$ 0.10US$ 0.10131,072no datano datano datano datano data66,175
Ministral 3 8B 2512Mistral AIopenUS$ 0.15US$ 0.15262,144no datano datano datano datano data93,502
Mistral LargeMistral AIclosedUS$ 2.00US$ 6.00128,00038.8%no datano datano datano datano data
Mistral Medium 3Mistral AIclosedUS$ 0.40US$ 2.00131,07259.5%no datano datano datano datano data
Mistral Medium 3.5Mistral AIundisclosedUS$ 1.50US$ 7.50262,144no datano datano datano datano datano data
Mistral Medium 3.1Mistral AIundisclosedUS$ 0.40US$ 2.00131,072no datano datano datano datano datano data
Mistral NemoMistral AIopenUS$ 0.029US$ 0.03131,07229.9%no datano datano datano data96,932
SabaMistral AIundisclosedUS$ 0.20US$ 0.6032,768no datano datano datano datano datano data
Mistral Small 3.1 24BMistral AIopenUS$ 0.35US$ 0.56128,000no datano datano datano datano data161,294
Mistral Small 3.2 24BMistral AIopenUS$ 0.10US$ 0.30256,000no datano datano datano datano data104,505
Mixtral 8x22B InstructMistral AIopenUS$ 2.00US$ 6.0065,536no datano datano datano datano data39,252
Voxtral Small 24B 2507Mistral AIopenUS$ 0.10US$ 0.3032,768no datano datano datano datano data6,535
Kimi K2 0711Moonshot AIopenUS$ 0.57US$ 2.30131,072no datano datano datano datano data277,844
Kimi K2 ThinkingMoonshot AIopenUS$ 0.60US$ 2.50262,14484.2%no datano datano datano data53,302
Nemotron 3 SuperNVIDIAopenUS$ 0.075US$ 0.40262,144no datano datano datano datano data35,797
GPT-3.5 Turbo (older v0613)OpenAIundisclosedUS$ 1.00US$ 2.004,095no datano datano datano datano datano data
GPT-3.5 Turbo 16kOpenAIundisclosedUS$ 3.00US$ 4.0016,385no datano datano datano datano datano data
GPT-4OpenAIclosedUS$ 30.00US$ 60.008,19135.7%no datano datano datano datano data
GPT-5 ProOpenAIclosedUS$ 15.00US$ 120.00400,000no datano datano datano datano datano data
GPT-5.1-CodexOpenAIundisclosedUS$ 1.25US$ 10.00400,000no datano datano datano datano datano data
GPT-5.1-Codex-MaxOpenAIundisclosedUS$ 1.25US$ 10.00400,000no datano datano datano datano datano data
GPT-5.1-Codex-MiniOpenAIundisclosedUS$ 0.25US$ 2.00400,000no datano datano datano datano datano data
GPT-5.2 ChatOpenAIundisclosedUS$ 1.75US$ 14.00128,000no datano datano datano datano datano data
GPT-5.2-CodexOpenAIundisclosedUS$ 1.75US$ 14.00400,000no datano datano datano datano datano data
GPT-5.2 ProOpenAIclosedUS$ 21.00US$ 168.00400,000no datano datano datano datano datano data
GPT-5.4 ProOpenAIclosedUS$ 30.00US$ 180.001,050,00094.6%no datano data46.3%no datano data
GPT-5.5 ProOpenAIclosedUS$ 30.00US$ 180.001,050,00093.9%no datano datano datano datano data
GPT-5.6 Terra ProOpenAIundisclosedUS$ 2.00US$ 12.001,050,000no datano datano datano datano datano data
GPT-6 Sol ProOpenAIundisclosedUS$ 2.00US$ 10.001,050,000no datano datano datano datano datano data
GPT-6.1 Sol ProOpenAIundisclosedUS$ 2.00US$ 10.001,050,000no datano datano datano datano datano data
GPT-6.1 Terra ProOpenAIundisclosedUS$ 1.00US$ 5.001,050,000no datano datano datano datano datano data
GPT AudioOpenAIundisclosedUS$ 2.50US$ 10.00128,000no datano datano datano datano datano data
GPT Audio MiniOpenAIundisclosedUS$ 0.60US$ 2.40128,000no datano datano datano datano datano data
GPT Chat LatestOpenAIundisclosedUS$ 5.00US$ 30.00400,000no datano datano datano datano datano data
gpt-oss-safeguard-20bOpenAIopenUS$ 0.075US$ 0.30131,072no datano datano datano datano datano data
o3OpenAIclosedUS$ 2.00US$ 8.00200,00081.8%62.3%no data49.4%no datano data
o3 ProOpenAIundisclosedUS$ 20.00US$ 80.00200,000no datano datano datano datano datano data
Qwen2.5 7B InstructAlibabaopenno datano datano datano datano datano datano datano data6,075,622
Qwen-PlusAlibabaclosedUS$ 0.26US$ 0.781,000,00048.1%no datano datano datano datano data
Qwen3 Coder FlashAlibabaundisclosedUS$ 0.20US$ 0.971,000,000no datano datano datano datano datano data
Qwen3 VL 30B A3B ThinkingAlibabaopenUS$ 0.29US$ 1.00262,144no datano datano datano datano data9,400
Qwen3.5-27BAlibabaopenUS$ 0.26US$ 2.60262,144no datano datano datano datano data1,661,103
Qwen3.5-FlashAlibabaundisclosedUS$ 0.065US$ 0.261,000,00082.3%no datano data20.3%no datano data
Qwen3.5 Plus 2026-02-15AlibabaundisclosedUS$ 0.26US$ 1.561,000,000no datano datano datano datano datano data
Qwen3.5 Plus 2026-04-20AlibabaundisclosedUS$ 0.30US$ 1.801,000,000no datano datano datano datano datano data
Qwen3.6 FlashAlibabaundisclosedUS$ 0.19US$ 1.121,000,00083.3%no datano data15.9%no datano data
Qwen3.6 PlusAlibabaclosedUS$ 0.33US$ 1.951,000,00088.4%57.9%no data44.1%no datano data
Qwen3.8 Max PrimeAlibabaundisclosedUS$ 4.00US$ 12.001,000,000no datano datano datano datano datano data
Qwen3.8 Omni FlashAlibabaundisclosedUS$ 0.15US$ 0.471,000,000no datano datano datano datano datano data
Step 3.5 FlashStepFunopenUS$ 0.10US$ 0.30262,144no datano datano datano datano data188,228
Step 3.7 FlashStepFunopenUS$ 0.20US$ 1.15262,144no datano datano datano datano data10,411
Hy3 previewTencentopenUS$ 0.18US$ 0.60262,144no datano datano datano datano data69,777
Grok 4.20xAIclosedUS$ 1.25US$ 2.502,000,00089.3%no datano data30.2%no datano data
Grok 4.3xAIundisclosedUS$ 1.25US$ 2.501,000,000no datano datano datano datano datano data
Grok 4.5xAIclosedUS$ 2.00US$ 6.00500,00093.4%no datano data48.3%no datano data
Grok Build 0.1xAIundisclosedUS$ 1.00US$ 2.00256,000no datano datano datano datano datano data
MiMo-V2.6-Pro-UltraSpeedXiaomiundisclosedUS$ 4.35US$ 8.701,048,576no datano datano datano datano datano data
GLM 4.5Zhipu AI (Z.ai)openUS$ 0.60US$ 2.20131,072no datano datano datano datano data113,522
GLM 4.5 AirZhipu AI (Z.ai)openUS$ 0.20US$ 1.10131,072no datano datano datano datano data1,258,503
GLM 4.5VZhipu AI (Z.ai)openUS$ 0.60US$ 1.8065,536no datano datano datano datano data15,564
GLM 5 TurboZhipu AI (Z.ai)undisclosedUS$ 1.20US$ 4.00202,752no datano datano datano datano datano data
GLM 5.3 FlashXZhipu AI (Z.ai)undisclosedUS$ 0.37US$ 1.251,048,576no datano datano datano datano datano data
GLM 5.3 FlashX LegalZhipu AI (Z.ai)undisclosedUS$ 0.37US$ 1.251,048,576no datano datano datano datano datano data
GLM 5.3 PrimeZhipu AI (Z.ai)undisclosedUS$ 2.80US$ 8.801,000,000no datano datano datano datano datano data
GLM 5V TurboZhipu AI (Z.ai)undisclosedUS$ 1.20US$ 4.00202,752no datano datano datano datano datano data

Which model to use for what

This pick follows the data and changes when the data does.

Agents

τ-bench Verified (airline) ?

What does this mean?

What it measures: τ-bench Verified (airline), on the OpenRouter Benchmarks leaderboard (CC BY 4.0, latest version we checked), simulates an airline support agent that uses systems and must follow company policy until the end of the conversation. The score is the share of tasks completed within the rules.

When it matters: It matters for agents that run tasks with tools. It shows whether the model follows policy across several steps, like an assistant that looks up an internal rule and finishes the request without inventing an exception.

Compare with

Text · Code · Image · Video · Voice · Transcription

Sources

  • Epoch AI, CC BY 4.0, latest version we checked.
  • OpenRouter Benchmarks, CC BY 4.0, latest version we checked.
  • OpenRouter, dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.
  • Hugging Face Hub, metadados públicos do Hub; licença de cada modelo no repositório, latest version we checked.

Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.