LMArcade - mini-swe-agent: leaderboard

LMArcade head-to-head human-voted playable-game Elo leaderboard inside the mini-swe-agent coding-agent harness (builds >= 5).

Metric: Elo Rating. Source: www.lmarcade.com. Saturation forecast: Around September 2027. 25 models tracked. Updated 22 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Fable 51676#32
2Claude Opus 4.71673#86
3Claude Opus 51669#26
4GPT-5.5Medium1660#56
5Claude Opus 4.81658#64
6Claude Sonnet 4.61643#179
7Claude Sonnet 51622#124
8Kimi K31573#62
9DeepSeek V4 Pro1570#201
10DeepSeek V4 Flash1527#273
11Qwen 3.5 397B A17B1516#270
12Kimi K2.61513#197
13GLM-5.11513#200
14GPT-5.4Medium1503#138
15Gemma 4 31B1483#390

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=lmarcade-mini-swe-agent · How It Works · Data refreshed daily, snapshot 2026-11-22.