Toolathlon
A tool-use benchmark focused on selecting, sequencing, and completing tasks with external tools.
Top models on Toolathlon — November 13, 2026
As of November 13, 2026, Muse Spark 1.1 leads the Toolathlon leaderboard with 75.6% , followed by Claude Opus 4.8 (59.9%) and GPT-5.6 Sol (58%).
Muse Spark 1.1
Meta
Reported by MetaClaude Opus 4.8
Anthropic
Reported by AnthropicGPT-5.6 Sol
OpenAI
Reported by OpenAI26 modelsAgentic3% of Agentic reference weightCurrentUpdated November 13, 2026
Leaderboard (26 models)
None of the 26 results on this page has a documented run by the benchmark or an independent evaluator: 24 reported by the labs that built the models and 2 reported in another publisher’s document.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Muse Spark 1.1MetaReported by Meta | 75.6% | Reported by Meta | Not reported | Closed |
| 2 | Claude Opus 4.8AnthropicReported by Anthropic | 59.9% | Reported by Anthropic | Not reported | Closed |
| 3 | GPT-5.6 SolOpenAIReported by OpenAI | 58% | Reported by OpenAI | Not reported | Closed |
| 4 | Gemini 3.5 FlashGoogleReported by Google | 56.5% | Reported by Google | Not reported | Closed |
| 5 | GPT-5.5OpenAIReported by OpenAI | 55.6% | Reported by OpenAI | Not reported | Closed |
| 6 | GPT-5.4OpenAIReported by OpenAI | 54.6% | Reported by OpenAI | Not reported | Closed |
| 7 | GPT-5.6 LunaOpenAIReported by OpenAI | 53.4% | Reported by OpenAI | Not reported | Closed |
| 8 | GPT-5.6 TerraOpenAIReported by OpenAI | 53.1% | Reported by OpenAI | Not reported | Closed |
| 9 | DeepSeek V4 Pro 0813DeepSeekReported by DeepSeek | 51.8% | Reported by DeepSeek | Not reported | Open |
| 10 | Kimi K2.6Moonshot AIReported by Moonshot AI | 50% | Reported by Moonshot AI | Not reported | Open |
| 11 | Step 3.7 FlashStepFunReported by StepFun | 49.5% | Reported by StepFun | Not reported | Open |
| 12 | DeepSeek V4 Pro (High)DeepSeekReported by DeepSeek | 49% | Reported by DeepSeek | Not reported | Open |
| 13 | GLM-5.2Z.AIReported by Z.AI | 48.2% | Reported by Z.AI | Not reported | Open |
| 14 | DeepSeek V4 Flash 0731DeepSeekReported by DeepSeek | 47.8% | Reported by DeepSeek | Not reported | Open |
| 15 | MiniMax M2.7MiniMaxReported by MiniMax | 46.3% | Reported by MiniMax | Not reported | Open |
| 16 | DeepSeek V4 ProDeepSeekReported by DeepSeek | 46.3% | Reported by DeepSeek | Not reported | Open |
| 17 | Claude Opus 4.5AnthropicAnother lab’s comparison table | 43.5% | Another lab’s comparison table | Not reported | Closed |
| 18 | DeepSeek V4 Flash (High)DeepSeekReported by DeepSeek | 43.5% | Reported by DeepSeek | Not reported | Open |
| 19 | GPT-5.4 miniOpenAIReported by OpenAI | 42.9% | Reported by OpenAI | Not reported | Closed |
| 20 | DeepSeek V4 FlashDeepSeekReported by DeepSeek | 40.7% | Reported by DeepSeek | Not reported | Open |
| 21 | Qwen3.6 PlusAlibabaReported by Alibaba | 39.8% | Reported by Alibaba | Not reported | Closed |
| 22 | GLM-5Z.AIAnother lab’s comparison table | 38% | Another lab’s comparison table | Not reported | Open |
| 23 | Qwen3.5 397BAlibabaReported by Alibaba | 36.3% | Reported by Alibaba | Not reported | Open |
| 24 | GPT-5.4 nanoOpenAIReported by OpenAI | 35.5% | Reported by OpenAI | Not reported | Closed |
| 25 | Kimi K2.5Moonshot AIReported by Moonshot AI | 27.8% | Reported by Moonshot AI | Not reported | Open |
| 26 | Qwen3.6-35B-A3BAlibabaReported by Alibaba | 26.9% | Reported by Alibaba | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
According to BenchLM.ai, Muse Spark 1.1 leads the Toolathlon benchmark with a score of 75.6%, followed by Claude Opus 4.8 (59.9%) and GPT-5.6 Sol (58%). There is significant spread across the leaderboard, making this benchmark effective at differentiating model capabilities.
26 models have been evaluated on Toolathlon. The benchmark falls in the Agentic category. BenchAlign v5.8 gives Toolathlon 3% of the Agentic reference weight, so it moves the Agentic leaderboard and the overall ranking. Reference weights are relative weights in the calibrated model, not fixed shares of a score.
About Toolathlon
Year
2026
Tasks
Multi-tool workflows
Format
Interactive tool-calling evaluation
Difficulty
Advanced tool use
Toolathlon is useful for judging whether a model can do more than answer in chat and instead complete multi-step tool workflows.
Freshness and provenance
Version
Toolathlon 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does Toolathlon measure?
A tool-use benchmark focused on selecting, sequencing, and completing tasks with external tools.
Which model scores highest on Toolathlon?
Muse Spark 1.1 by Meta currently leads with a score of 75.6% on Toolathlon.
How many models are evaluated on Toolathlon?
26 AI models have published results on Toolathlon in the BenchLM catalog.
Compare top models on Toolathlon
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.