MERA - USE: leaderboard
MERA is the Russian open benchmark run by the AI Alliance Russia; this is its USE task: problems from the Russian Unified State Exam, scored on the exam’s own grading scheme. Each MERA task keeps its own native metric, so the boards are not on one shared scale.
Metric: Grade, normalized (%). Source: mera.a-ai.ru. Saturation forecast: Around January 2027. 632 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 89.02 | #63 |
| 2 | GPT-5.4 | 85.98 | #79 |
| 3 | GLM-5.1 | 77.75 | #97 |
| 4 | Claude Opus 4.5 | 73.73 | #83 |
| 5 | Qwen 3.5 122B A10B | 71.47 | #173 |
| 6 | Qwen 3.5 27B | 70.49 | #187 |
| 7 | Kimi K2.5 | 68.92 | #144 |
| 8 | Qwen 3 235B A22B 2507 (Thinking) | 68.63 | #260 (Qwen 3 235B A22B 2507) |
| 9 | DeepSeek V4 Pro | 67.55 | #102 |
| 10 | DeepSeek V4 Flash | 67.55 | #137 |
| 11 | Qwen 3.6 35B A3B | 66.96 | #226 |
| 12 | Qwen 3.6 27B | 66.67 | #170 |
| 13 | Gemma 4 31B (IT) | 61.96 | #176 |
| 14 | Qwen 3.5 35B A3B | 61.18 | #256 |
| 15 | GigaChat 2 Max | 58.14 | #536 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-use · How It Works · Data refreshed daily, snapshot 2026-10-20.