MERA - MaMuRAMu: leaderboard
MERA is the Russian open benchmark run by the AI Alliance Russia; this is its MaMuRAMu task: a massive multitask Russian exam battery in the MMLU mould, built natively rather than translated. Each MERA task keeps its own native metric, so the boards are not on one shared scale.
Metric: Accuracy (%). Source: mera.a-ai.ru. Saturation forecast: Estimated already saturated. 632 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 92.61 | #63 |
| 2 | GPT-5.4 | 92.02 | #79 |
| 3 | Claude Opus 4.5 | 92 | #83 |
| 4 | GLM-5.1 | 91.38 | #97 |
| 5 | GPT-5.2 | 90.3 | #110 |
| 6 | Qwen 3.5 122B A10B | 90.21 | #173 |
| 7 | Kimi K2.5 | 90.04 | #144 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 89.41 | #260 (Qwen 3 235B A22B 2507) |
| 9 | Qwen 3.5 27B | 89.34 | #187 |
| 10 | Claude 3.7 Sonnet | 89.03 | #249 |
| 11 | Qwen 3.6 27B | 88.75 | #170 |
| 12 | Qwen 3.6 35B A3B | 88.75 | #226 |
| 13 | Qwen 3.5 35B A3B | 88.32 | #256 |
| 14 | DeepSeek V3 | 88.18 | #321 |
| 15 | Gemini 2.0 Flash | 87.81 | #342 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-mamuramu · How It Works · Data refreshed daily, snapshot 2026-10-20.