MERA Code - ruCodeEval: leaderboard
MERA Code (Russian) ruCodeEval task: Russian-prompted HumanEval-style generation, scored by unit tests. 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: pass@1 (%). Source: mera.a-ai.ru. Saturation forecast: Not forecast. 66 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Next 80B A3B (Thinking) | 76.22 | #320 (Qwen 3 Next 80B A3B) |
| 2 | DeepSeek V4 Pro | 75.79 | #102 |
| 3 | Qwen 3 30B A3B 2507 (Thinking) | 74.15 | #376 (Qwen 3 30B A3B 2507) |
| 4 | Qwen 3 4B 2507 (Thinking) | 72.62 | #538 (Qwen 3 4B 2507) |
| 5 | GLM-5.1 | 71.65 | #97 |
| 6 | Qwen 3 14B | 67.38 | #537 |
| 7 | Qwen 3 32B | 66.95 | #443 |
| 8 | Qwen 3 8B | 63.23 | #674 |
| 9 | GPT-5.1 Codex Max | 63.23 | #100 |
| 10 | Gemini 2.5 Flash | 61.04 | #245 |
| 11 | Qwen 3 4B | 58.05 | #829 |
| 12 | MiniMax-M2.7 | 54.7 | #258 |
| 13 | Claude Opus 4.6 | 53.6 | #63 |
| 14 | GigaChat 2 Max | 53.41 | #536 |
| 15 | DeepSeek V4 Flash | 52.87 | #137 |
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-rucodeeval · How It Works · Data refreshed daily, snapshot 2026-10-20.