METR Q4 2026 Lifecycle Cost Horizon - TH1.1 + TH2 (50%): leaderboard
METR's Q4 2026 capability update (20 November 2026): the task cost, in US dollars, at which a model succeeds half the time on the pooled TH1.1 + TH2 tasks. It replaces human time on the usual time-horizon axis with lifecycle cost: Claude Opus 5.5's estimate of what it costs to build a task and then run and score it 100 times. Unbounded; higher is better.
Metric: 50% Lifecycle Cost Horizon (USD). Source: metr.org. Saturation forecast: Around December 2027. 15 models tracked. Updated 21 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 34602 | #9 |
| 2 | GPT-6 Astra | 30763 | #10 |
| 3 | Claude Fable 5.1 | 20783 | #20 |
| 4 | Claude Opus 5 | 17988 | #26 |
| 5 | Claude Fable 5 | 15688 | #33 |
| 6 | GPT-5.6 Sol | 14060 | #37 |
| 7 | GPT-5.5 | 12802 | #54 |
| 8 | Claude Opus 4.8 | 11995 | #65 |
| 9 | Claude Opus 4.7 | 8246 | #85 |
| 10 | Claude Opus 4.6 | 6850 | #113 |
| 11 | GPT-5.4 | 6368 | #135 |
| 12 | Claude Opus 4.5 | 5036 | #160 |
| 13 | GPT-5 | 4214 | #178 |
| 14 | O1 | 1568 | #403 |
| 15 | GPT-4o | 678 | #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-lifecycle-cost-horizon-th1-1-plus-th2-50 · How It Works · Data refreshed daily, snapshot 2026-11-21.