MERA Code - CodeCorrectness: leaderboard
MERA Code (Russian) CodeCorrectness task: judging whether a snippet is correct, rather than writing one. Each task keeps its own metric — pass rates, CodeBLEU and chrF all appear across this suite — so read the metric label before comparing boards.
Metric: Accuracy (%). Source: mera.a-ai.ru. Saturation forecast: Not forecast. 66 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 94.27 | #79 |
| 2 | GPT-5.1 Codex Max | 91.33 | #100 |
| 3 | Kimi K2.5 | 89.64 | #144 |
| 4 | Qwen 3.6 27B | 87.44 | #170 |
| 5 | GLM-5.1 | 86.77 | #97 |
| 6 | Claude Opus 4.6 | 85.89 | #63 |
| 7 | Qwen 3.6 35B A3B | 84.94 | #226 |
| 8 | LFM2.5-1.2B Instruct | 84.5 | #1462 |
| 9 | Qwen 3 32B | 84.13 | #443 |
| 10 | Qwen 2.5 1.5B Instruct | 83.69 | #1407 |
| 11 | Qwen 3 30B A3B 2507 (Thinking) | 82.95 | #376 (Qwen 3 30B A3B 2507) |
| 12 | Qwen 2.5 0.5B Instruct | 82.73 | #1529 |
| 13 | LFM2 8B A1B | 82.59 | #1263 |
| 14 | OLMo 3 7B Instruct | 81.56 | #1218 |
| 15 | LFM2 24B A2B | 81.26 | #1379 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mera-code-codecorrectness · How It Works · Data refreshed daily, snapshot 2026-10-20.