EgoBench - Dynamic Easy Mode - Micro Accuracy: leaderboard

Metric: MicroAcc (%; share of ground-truth tool calls issued, pooled over tasks, cooperative simulated user revealing one step per turn). Source: arxiv.org. Saturation forecast: Around July 2027. 8 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)44.42
2Qwen 3.5 397B A17B (Non-reasoning)41.66
3Seed 2.0 Pro (Non-reasoning)38.69
4Qwen 3.6 Plus (Non-reasoning)36.58
5Kimi K2.5 (Non-reasoning)35.42
6GLM-5V Turbo29.7
7MiMo-V2-Omni27.99
8Qwen 3 VL 235B A22B Instruct15.02

Interactive version: theaggregate.ai/benchmark?slug=egobench-dynamic-easy-mode-micro-accuracy · How It Works · Data refreshed daily, snapshot 2026-09-25.