ToolathlonBrowse 296

Toolathlon

Can the model plan, use tools, and finish a multi-step task?

Version not specifiedExact reported variant
Agents11 ranked models12 reported values2 reportsHigher is better

Top rankings

One row per model, using its best reported score across effort settings.

11
RankModelBest scoreBest reported setting
1Claude Fable 5Anthropic61.7%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
2Claude Mythos 5Anthropic61.7%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
3Claude Mythos PreviewAnthropic61.1%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
4Claude Opus 4.8Anthropic59.9%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
5GPT-5.6 SolOpenAI58.0%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
6Gemini 3.5 FlashGoogle56.5%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
7GPT-5.5OpenAI55.6%Reported configurationOpenAI report2 values · 2 reportsJul 9, 2026 · source
8GPT-5.6 LunaOpenAI53.4%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
9GPT-5.6 TerraOpenAI53.1%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source
10Gemini 3 FlashGoogle49.4%Reported configurationGoogle report1 value · 1 reportMay 19, 2026 · source
11Gemini 3.1 Pro PreviewGoogle48.8%Reported configurationOpenAI report1 value · 1 reportJul 9, 2026 · source

Effort curve

Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.

0
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology

Can the model plan, use tools, and finish a multi-step task?

Agentic execution across tools, environments, or long-running workflows. Higher is better. The value is the percentage reported in this lab's table.

Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.

Method / source