METR Q4 2026 TH2 - Reverse Conway (Offline): leaderboard

METR's Q4 2026 capability update (20 November 2026): mean score from 0 to 1 on one task of TH2, METR's new long-task suite. The Reverse Conway task (find starting boards that evolve into target patterns in Conway's Game of Life) in the variant METR's data labels offline. Higher is better.

Metric: Mean Score (0-1). Source: metr.org. Saturation forecast: Around April 2028. 15 models tracked. Updated 21 Nov 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 5.50.2067#9
2GPT-6 Astra0.1494#10
3Claude Fable 5.10.103#20
4Claude Fable 50.0874#33
5Claude Opus 50.0783#26
6GPT-5.6 Sol0.0655#37
7Claude Opus 4.80.0654#65
8Claude Opus 4.70.0525#85
9GPT-5.50.0512#54
10Claude Opus 4.50.0437#160
11Claude Opus 4.60.0297#113
12GPT-5.40.0217#135
13GPT-4o0#751
14GPT-50#178
15O10#403

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=metr-q4-2026-th2-reverse-conway-offline · How It Works · Data refreshed daily, snapshot 2026-11-21.