METR time horizons leaderboard
- Claude Opus 4.6 leads
- 78.9%
- models measured
- 21
- lead over #2
- +1.8 pts
- median
- 65.4%
Leader on METR time horizons · read October 21
On October 21, 2026, 26 of our catalog's models had a METR time horizons result, and Anthropic Claude Opus 4.6 led with 78.9%, ahead of Google Gemini 3.1 Pro Preview at 77.0%.
METR measures how long a software task, timed on human experts, a model can finish with 50% reliability. Epoch AI publishes the score it derives from METR results.
Role in the index
METR time horizons is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Opus 4.6Anthropic | 78.9% | 57.2 | $25.00 | |
| 2 | Gemini 3.1 Pro PreviewGoogle | 77.0% | 32.3 | $12.00 | |
| 3 | GPT-5.2OpenAI | 75.3% | 55.9 | $14.00 | |
| 4 | Claude Opus 4.5Anthropic | 75.0% | 44.0 | $25.00 | |
| 5 | GPT-5.3-CodexOpenAI | 74.5% | 60.0 | $14.00 | |
| 6 | GPT-5.4OpenAI | 74.3% | 64.4 | $15.00 | |
| 7 | GPT-5.1-Codex-MaxOpenAI | 70.8% | Not benchmarked | $10.00 | |
| 8 | GPT-5OpenAI | 69.6% | 38.8 | $10.00 | |
| 9 | Claude Sonnet 4.5Anthropic | 67.4% | 39.5 | $15.00 | |
| 10 | Claude Opus 4.1Anthropic | 66.8% | Not benchmarked | $75.00 | |
| 11 | o3OpenAI | 65.4% | Not benchmarked | $8.00 | |
| 12 | Claude Opus 4RetiredAnthropic | 63.9% | Not benchmarked | $75.00 | |
| 13 | o4 MiniOpenAI | 63.9% | Not benchmarked | $4.40 | |
| 14 | Claude 3.7 SonnetRetiredAnthropic | 60.0% | Not benchmarked | $15.00 | |
| 15 | Kimi K2 ThinkingMoonshot AI | 59.2% | Not benchmarked | $2.50 | |
| 16 | gpt-oss-120bOpenAI | 56.6% | Not benchmarked | $0.50 | |
| 17 | Gemini 2.5 Pro Preview 06-05Google | 55.4% | Not benchmarked | $10.00 | |
| 18 | R1 0528DeepSeek | 53.8% | Not benchmarked | $2.18 | |
| 19 | R1RetiredDeepSeek | 51.9% | Not benchmarked | $2.50 | |
| 20 | o1OpenAI | 51.1% | Not benchmarked | $60.00 | |
| 21 | DeepSeek V3 0324DeepSeek | 49.6% | Not benchmarked | $1.07 | |
| 22 | Claude 3.5 SonnetRetiredAnthropic | 45.2% | Not benchmarked | $15.00 | |
| 23 | GPT-4o (2024-05-13)OpenAI | 40.8% | Not benchmarked | $15.00 | |
| 24 | Qwen2.5 72B InstructQwen | 35.8% | Not benchmarked | $0.40 | |
| 25 | Claude 3 OpusRetiredAnthropic | 29.5% | Not benchmarked | $75.00 | |
| 26 | GPT-3.5 Turbo InstructOpenAI | 21.5% | Not benchmarked | $2.00 |
Only models in our catalog are listed, and each appears once with its best published run. 215 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.