MacAgentBench - Multi-App: leaderboard

Metric: Pass@1 (%; share of the 140 Multi-App tasks solved on a single attempt, deterministic rule-based checks; the benchmark's baseline pure-GUI agent (screenshots and mouse/keyboard actions only), at most 50 steps). Source: arxiv.org. Saturation forecast: Around June 2027. 16 models tracked.

Top models

#ModelScore
1GPT-5.448.6
2Claude Opus 4.620.7
3Gemini 3.1 Pro (Preview)13.6
4Qwen 3 VL 32B (Thinking)7.9
5Qwen 3 VL 235B A22B6.4
6Qwen 3 VL 8B (Thinking)2.1
7uitars-1.5-7B1.4

Interactive version: theaggregate.ai/benchmark?slug=macagentbench-multi-app · How It Works · Data refreshed daily, snapshot 2026-09-26.