METR Q4 2026 TH2 - Earthborne Rangers (Cheesable): 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 plays Earthborne Rangers, a card game of managing resources, lightly adapted from Epoch's EBR-Bench; this is the variant METR's data labels cheesable. The score is the share of 21 objectives completed. Higher is better.

Metric: Mean Score (0-1). Source: metr.org. Saturation forecast: Around February 2027. 14 models tracked. Updated 21 Nov 2026.

Top models

#ModelScoreOverall rank
1GPT-6 Astra0.8#10
2Claude Opus 5.50.7238#9
3Claude Opus 50.5524#26
4Claude Fable 5.10.4667#20
5GPT-5.6 Sol0.4381#37
6GPT-5.50.4286#54
7GPT-5.40.3524#135
8Claude Opus 4.60.2952#113
9Claude Opus 4.80.2952#65
10Claude Opus 4.50.2857#160
11Claude Fable 50.2762#33
12Claude Opus 4.70.2667#85
13GPT-50.2571#178
14O10.1238#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-earthborne-rangers-cheesable · How It Works · Data refreshed daily, snapshot 2026-11-21.