OmniaBench - ToC: leaderboard

Metric: Pass@1 (%; tasks from the consumer-facing (ToC) 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 March 2027. 22 models tracked.

Top models

#ModelScore
1Claude Sonnet 5 (Max)60.47
2GPT-5.6 Sol (High)59.07
3GLM-5.2 (Max)55.81
4DeepSeek V4 Pro (Max)54.88
5GPT-5.5 (High)53.02
6DeepSeek V4 Flash (Max)51.63
7Claude Opus 4.7 (Max)51.63
8Kimi K2.6 (Thinking)49.3
9Qwen 3.7 Max (Thinking)48.84
10Gemini 3.5 Flash (High)46.98
11Gemini 3.1 Pro (Preview)39.53
12Qwen 3.6 27B38.6
13GLM-5 (Thinking)37.67
14Qwen 3.5 397B A17B35.35
15Qwen 3.6 35B A3B31.63

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