LLM Stats (MultiPL-E): leaderboard
Code generation across 18+ programming languages translated from HumanEval and MBPP, measuring multilingual coding capability.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 235B A22B 2507 Instruct | 87.9 | #300 |
| 2 | Qwen 3 Next 80B A3B Instruct | 87.8 | #368 |
| 3 | Qwen 3 VL 235B A22B Instruct | 86.1 | #273 |
| 4 | Kimi K2 | 85.7 | #243 |
| 5 | Kimi K2 0905 | 85.7 | #288 |
| 6 | Qwen 2.5 32B Instruct | 75.4 | #498 |
| 7 | Qwen 2.5 72B Instruct | 75.1 | #454 |
| 8 | Qwen 2.5 14B Instruct | 72.8 | #647 |
| 9 | Qwen 2.5 7B Instruct | 70.4 | #862 |
| 10 | Qwen 2 72B Instruct | 69.2 | #576 |
| 11 | Qwen 3 235B A22B | 65.94 | #313 |
| 12 | Qwen2.5-Omni-7B | 65.8 | #624 |
| 13 | Qwen 2 7B Instruct | 59.1 | #991 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-multipl-e · How It Works · Data refreshed daily, snapshot 2026-10-20.