VECTOR WIREAI INTELLIGENCE
NVDA$1,847+3.2%MSFT$512+1.1%GOOGL$199-0.4%META$728+2.7%AMD$184-1.2%TSM$212+0.6%PLTR$98+4.1%AI IDX4,821+1.9%
PKT
SEEDRefresh Models Deals Regulatory Sources

toolathlon

43 models tracked

Toolathlon — Distinct named tool-use benchmark; sampling:pass@3 variant

#ModelVendorBest scoreRunsLast seen
1Muse Spark 1.1Meta75.612026-08-23
2Kimi K3Moonshot73.212026-08-23
3GLM-5.3Z.ai7322026-08-23
4Qwen3.8 Max PreviewAlibaba72.512026-08-23
5DeepSeek-V4-FlashDeepSeek70.352026-08-23
6Claude Fable 5Anthropic61.712026-06-30
7Claude Mythos 5Anthropic61.112026-06-30
8Claude Opus 4.8Anthropic59.922026-08-23
9Claude Opus 4.7Anthropic59.312026-06-30
10GPT-5.6 SolOpenAI5812026-08-23
11Claude Opus 4.6Anthropic56.862026-07-04
12Gemini 3.5 FlashGoogle56.532026-08-24
13GPT-5.5OpenAI55.612026-08-23
14GPT-5.4OpenAI54.662026-08-23
15Inkling-SmallThinking Machines54.412026-08-23
16Claude Sonnet 5Anthropic54.322026-08-23
17GPT-5.6 LunaOpenAI53.412026-08-23
18GPT-5.6 TerraOpenAI53.112026-08-23
19DeepSeek-V4-ProDeepSeek51.8102026-08-23
20Kimi K2.6Moonshot5042026-08-23
21laguna-s-2.1poolside49.712026-08-23
22Gemini 3 Flash PreviewGoogle49.422026-08-23
23Claude Sonnet 4.6Anthropic49.412026-06-30
24Gemini 3.1 ProGoogle48.862026-07-04
25Hy3-previewTencent48.512026-08-23
26GLM-5.2 Full Open SourceZ.ai48.212026-08-23
27MiniMax M2.7MiniMax46.312026-08-23
28GPT-5.2OpenAI46.322026-08-23
29MiniMax M2.1MiniMax43.522026-08-23
30Claude Opus 4.5Anthropic43.512026-06-05
31GPT-5.4 miniOpenAI42.912026-08-23
32Claude Sonnet 4.5Anthropic4122026-06-30
33ClaudesAnthropic4112026-08-12
34GLM-5.1Z.ai40.732026-08-23
35Qwen3.6 PlusAlibaba39.812026-08-23
36Qwen3.5 397B A17BAlibaba38.312026-08-23
37Gemini 3 ProGoogle36.412026-06-05
38GPT-5.4 nanoOpenAI35.512026-08-23
39DeepSeek-V3.2DeepSeek35.222026-08-23
40DeepSeek-V3.2-SpecialeDeepSeek35.212026-08-23
41Kimi K2.5Moonshot27.812026-05-03
42Qwen3.6 35B A3BAlibaba26.912026-08-23
43MiniMax M2MiniMax16.712026-06-05

Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.