MERA v2 - IFHardBench: leaderboard
MERA v2 is the second generation of the Russian MERA suite, rebuilt after the v1 tasks saturated; this board is hard instruction following, where the constraints are the difficulty.
Metric: Score (%). Source: mera.a-ai.ru. Saturation forecast: Around January 2027. 40 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.7 Flash | 95.8 | #23 |
| 2 | Grok 4.6 | 88.4 | #26 |
| 3 | GPT-5.6 Sol | 86.3 | #19 |
| 4 | Claude Opus 5 | 78.6 | #12 |
| 5 | Qwen 3.5 397B A17B | 73.2 | #146 |
| 6 | GPT-5.6 Terra | 67.1 | #41 |
| 7 | GPT-5.6 Luna | 65.8 | #68 |
| 8 | GigaChat 3.5 Reasoning | 62.8 | #167 |
| 9 | Grok 4.3 | 58.4 | #188 |
| 10 | MiMo-V2.5-Pro | 56.4 | #158 |
| 11 | Qwen 3.6 27B | 52.8 | #170 |
| 12 | MiniMax-M3 | 51.6 | #136 |
| 13 | Qwen 3.5 122B A10B | 48.7 | #173 |
| 14 | Solar Pro 4 | 47.5 | #215 |
| 15 | Qwen 3.5 27B | 46.6 | #187 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-v2-ifhardbench · How It Works · Data refreshed daily, snapshot 2026-10-20.