Benchmarks · math
OTIS Mock AIME 2024-2025
The OTIS Mock AIME problems (2024 and 2025), run by Epoch AI. Accuracy over repeated samples. Data: Epoch AI, CC BY.
163 models · 238 results · data as of 2026-10-29 · epoch.ai ↗
Leaderboard
Fable 5.1 max
All results
238 rows| # | Model | Effort | OTIS Mock AIME 2024-2025 | Harness |
|---|---|---|---|---|
| 1 | Fable 5.1 | max | 100.0% | Epoch AI |
| 2 | Fable 5 | high | 100.0% | Epoch AI |
| 3 | Opus 5.5 | max | 100.0% | Epoch AI |
| 4 | Sonnet 5.5 | max | 100.0% | Epoch AI |
| 5 | Gemini 4 Argon | high | 100.0% | Epoch AI |
| 6 | GPT-5.5 Pro | xhigh | 100.0% | Epoch AI |
| 7 | GPT-5.5 | xhigh | 100.0% | Epoch AI |
| 8 | GPT-5.6 Sol | max | 100.0% | Epoch AI |
| 9 | GPT-6 Astra | max | 100.0% | Epoch AI |
| 10 | GPT-6 Sol | max | 100.0% | Epoch AI |
| 11 | GPT-6.1 Sol | max | 100.0% | Epoch AI |
| 12 | GPT-6.1 Terra | max | 100.0% | Epoch AI |
| 13 | Qwen 3.8 Max | xhigh | 100.0% | Epoch AI |
| 14 | Fable 5 | max | 99.7% | Epoch AI |
| 15 | GPT-5.6 Terra | max | 99.7% | Epoch AI |
| 16 | DeepSeek V4.1 Flash 1017 | max | 99.4% | Epoch AI |
| 17 | Grok 4.6 | xhigh | 99.2% | Epoch AI |
| 18 | Muse Spark 1.3 | xhigh | 99.2% | Epoch AI |
| 19 | Haiku 5.5 | xhigh | 98.9% | Epoch AI |
| 20 | Gemini 3.8 Flash | high | 98.9% | Epoch AI |
| 21 | GPT-6 Luna | max | 98.9% | Epoch AI |
| 22 | DeepSeek V4 Pro 0813 | max | 98.6% | Epoch AI |
| 23 | Opus 4.8 | max | 98.3% | Epoch AI |
| 24 | DeepSeek V4.1 Flash | max | 98.3% | Epoch AI |
| 25 | GPT-5.6 Luna | max | 98.3% | Epoch AI |
| 26 | Mistral Large 4 | high | 98.3% | Epoch AI |
| 27 | Grok 4.7 | xhigh | 98.1% | Epoch AI |
| 28 | Opus 4.7 | xhigh | 97.8% | Epoch AI |
| 29 | Fable 5 | low | 97.8% | Epoch AI |
| 30 | Opus 4.8 | low | 97.8% | Epoch AI |