MERA - ruHateSpeech: leaderboard
MERA is the Russian open benchmark run by the AI Alliance Russia; this is its ruHateSpeech task: hate-speech detection in Russian. 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. 631 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Median Human | 98.5 | |
| 2 | GPT-5.4 | 94.72 | #79 |
| 3 | GLM-5.1 | 93.96 | #97 |
| 4 | Qwen 3.6 27B | 93.21 | #170 |
| 5 | Qwen 3.5 27B | 92.83 | #187 |
| 6 | Qwen 3.5 122B A10B | 92.83 | #173 |
| 7 | Qwen 3 235B A22B 2507 (Thinking) | 92.83 | #260 (Qwen 3 235B A22B 2507) |
| 8 | Qwen 3.6 35B A3B | 92.08 | #226 |
| 9 | Qwen 3 30B A3B 2507 (Thinking) | 92.08 | #376 (Qwen 3 30B A3B 2507) |
| 10 | Claude Opus 4.6 | 91.32 | #63 |
| 11 | Kimi K2.5 | 90.94 | #144 |
| 12 | GPT-5.2 | 90.57 | #110 |
| 13 | Qwen 3 32B | 90.19 | #443 |
| 14 | Claude Opus 4.5 | 90.19 | #83 |
| 15 | Qwen 3.5 35B A3B | 89.81 | #256 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-ruhatespeech · How It Works · Data refreshed daily, snapshot 2026-10-20.