MERA Reason - MMReD: leaderboard
MERA Reason is the Russian reasoning suite MERA launched on 2026-07-24, run as its own Hugging Face Space; this board is the MMReD multi-step reasoning set.
Metric: Score. Source: huggingface.co. Saturation forecast: Around April 2027. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 27B | 96.36 | #187 |
| 2 | DeepSeek V4 Pro | 94.94 | #102 |
| 3 | Claude Opus 4.8 | 93.97 | #37 |
| 4 | Qwen 3.5 122B A10B | 93.07 | #173 |
| 5 | GLM-5 | 93.01 | #141 |
| 6 | Qwen 3.6 27B | 91.96 | #170 |
| 7 | Gemma 4 31B (IT) | 91.86 | #176 |
| 8 | Qwen 3.5 35B A3B | 88.75 | #256 |
| 9 | Qwen 3.6 35B A3B | 87.41 | #226 |
| 10 | gemma-4-E4B-it | 85.61 | #566 |
| 11 | DeepSeek V4 Flash | 80.69 | #137 |
| 12 | Qwen 3.5 9B | 77.87 | #379 |
| 13 | Qwen 3 Next 80B A3B (Thinking) | 73.85 | #320 (Qwen 3 Next 80B A3B) |
| 14 | Qwen 3 4B 2507 (Thinking) | 72.07 | #538 (Qwen 3 4B 2507) |
| 15 | Qwen 3 30B A3B 2507 (Thinking) | 69.95 | #376 (Qwen 3 30B A3B 2507) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-reason-mmred · How It Works · Data refreshed daily, snapshot 2026-10-20.