Toolathlon
7 models tested · Updated 2026-06-09 · Verified sources only
Claude Mythos 5 leads at 61.7%
1
Anthropic · Blog/OpenAI · 2026-06-09
Beats GPT-5.6 Sol (58) on Toolathlon; matches Mythos Preview (61.1) and Fable 5 (61.7).
61.7%
2
Anthropic · Blog/Z.ai · 2026-06-13
Highest among compared models.
59.9%
3
Google DeepMind · Blog/Google DeepMind · 2026-06-10
Highest among compared models. Edges GPT-5.5 (55.6).
56.5%
4
OpenAI · Blog/OpenAI · 2026-04-23
Toolathlon evaluates tool-use proficiency across diverse real-world tasks.
55.6%
5
DeepSeek · HuggingFace/deepseek-ai · 2026-04-24
Strong tool-use for open-weights model. Trails GPT-5.5 by 4pp.
51.8%
6
StepFun · Blog/StepFun · 2026-05-29
Toolathlon. Multi-tool coordination.
49.5%
7
Z.ai · Blog/Z.ai · 2026-06-13
Up from GLM-5.1 (40.7). Agentic tool use.
48.2%