OmniaBench - ToE: leaderboard

Metric: Pass@1 (%; tasks from the employee-oriented (ToE) domain split; a task counts as solved only when every rubric item is satisfied (Program tasks by a VerifyCode check); up to 200 tool-call steps; challenging set of 644 tasks evaluated June 29 to July 15, 2026). Source: arxiv.org. Saturation forecast: Around November 2027. 22 models tracked.

Top models

#ModelScore
1GPT-5.5 (High)58.49
2GLM-5.2 (Max)57.55
3Qwen 3.7 Max (Thinking)55.66
4DeepSeek V4 Pro (Max)54.72
5DeepSeek V4 Flash (Max)52.83
6GPT-5.6 Sol (High)51.89
7Claude Opus 4.7 (Max)51.89
8Kimi K2.6 (Thinking)50.94
9Claude Sonnet 5 (Max)48.11
10Qwen 3.6 27B44.34
11Qwen 3.5 397B A17B43.4
12Gemini 3.1 Pro (Preview)42.45
13GLM-5 (Thinking)41.51
14Gemini 3.5 Flash (High)39.62
15Qwen 3.5 35B A3B36.79

Interactive version: theaggregate.ai/benchmark?slug=omniabench-toe · How It Works · Data refreshed daily, snapshot 2026-09-29.