METR Horizon AI Model Rankings
| # | Model | Horizon | Cost |
|---|---|---|---|
| 1 | Gemini 3.1 Pro Google | 6.4h | $4.50/M |
| 2 | GPT 5.2 OpenAI | 5.9h | $4.81/M |
| 3 | GPT 5.3 Codex OpenAI | 5.8h | $4.81/M |
| 4 | GPT 5.4 OpenAI | 5.7h | $5.63/M |
| 5 | Claude Opus 4.5 Anthropic | 4.9h | $10.00/M |
| 6 | Gemini 3 Pro Google | 3.7h | — |
| 7 | GPT 5 OpenAI | 3.4h | $3.44/M |
| 8 | GPT 5.1 Codex Max OpenAI | 2.7h | $3.44/M |
| 9 | Claude Opus 4.1 Anthropic | 1.7h | $30.00/M |
| 10 | o3 OpenAI | 1.5h | $3.50/M |
| 11 | o4-mini (High) OpenAI | 1.3h | $1.93/M |
| 12 | Kimi K2 MoonshotAI | 54min | $1.00/M |
| 13 | GPT Mini (120b) OpenAI | 42min | $0.07/M |
| 14 | o1 OpenAI | 39min | $26.25/M |
| 15 | DeepSeek R1 (0528) DeepSeek | 31min | $0.91/M |
| 16 | Claude 3.5 Sonnet Anthropic | 30min | — |
| 17 | DeepSeek R1 DeepSeek | 27min | $1.15/M |
| 18 | DeepSeek V3 (0324) DeepSeek | 23min | $0.34/M |
| 19 | DeepSeek V3 DeepSeek | 18min | $0.35/M |
| 20 | GPT 4o OpenAI | 9min | $4.38/M |
| 21 | GPT 4 Turbo OpenAI | 7min | $15.00/M |
| 22 | Qwen2.5 72B Qwen | 5min | $0.37/M |
| 23 | GPT 3.5 Turbo Instruct OpenAI | 1min | $1.63/M |
23 of 23 models, ranked by METR Time Horizon