METR Q4 2026 TH2 - LMCA: 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 agent rates critiques of philosophical arguments and is scored on agreement with expert ratings (dataset from Cooper et al., including its held-out set). METR estimates experts reach 0.5 in 5 minutes and 0.8 in 15. Higher is better.

Metric: Mean Score (0-1). Source: metr.org. Saturation forecast: Around February 2027. 15 models tracked. Updated 21 Nov 2026.

Top models

#ModelScoreOverall rank
1Claude Opus 5.50.6899#9
2Claude Fable 5.10.6822#20
3Claude Opus 50.6817#26
4Claude Fable 50.6514#33
5GPT-6 Astra0.6454#10
6Claude Opus 4.80.6304#65
7GPT-5.50.5747#54
8GPT-5.6 Sol0.574#37
9Claude Opus 4.70.5723#85
10Claude Opus 4.60.5588#113
11GPT-5.40.5207#135
12Claude Opus 4.50.4389#160
13GPT-50.389#178
14O10.2558#403
15GPT-4o0.0973#751

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-lmca · How It Works · Data refreshed daily, snapshot 2026-11-21.