Inference
EN

Toolathlon

A tool-use benchmark focused on selecting, sequencing, and completing tasks with external tools.

ModelsModels with a result.22
Top resultBest result on this benchmark.75.6%Muse Spark 1.1
Top-3 spreadPoints from first to third.17.6 pts
YearYear of release.—

Result and price

22 of 22 results

Results

22 results
#ModelResultThe score from the source.
01Muse Spark 1.1Above fitFar above this model's index.75.6%
02Claude Opus 4.859.9%
03GPT-5.6 Sol58.0%
04Gemini 3.5 Flash56.5%
05GPT-5.555.6%
06GPT-5.454.6%
07GPT-5.6 Luna53.4%
08GPT-5.6 Terra53.1%
09DeepSeek V4 Pro 081351.8%
10Kimi K2.650.0%
11Step 3.7 Flash49.5%
12GLM-5.248.2%
13DeepSeek V4 Flash 073147.8%
14MiniMax M2.746.3%
15Claude Opus 4.543.5%
16GPT-5.4 mini42.9%
17Qwen3.6 Plus39.8%
18GLM-538.0%
19Qwen3.5 397B36.3%
20GPT-5.4 nano35.5%
21Kimi K2.527.8%
22Qwen3.6-35B-A3B26.9%