BLXBench pass rate
Code · 2026-04-25
Pass-rate component of the BLXBench public leaderboard. It reports the percentage of benchmark tests passed by each model entry.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.5 | 61.9 |
| GPT-5.3-Codex | 61.2 |
| Kimi K2.6 | 60.6 |
| Opus 4.7 | 60.5 |
| Opus 4.8 | 60.1 |
| MiniMax M3 | 58.0 |
| Claude Fable 5 | 56.4 |
| MiMo-V2.5 | 55.3 |
| Qwen3.7-Max | 54.9 |
| MiMo-V2.5-Pro | 54.0 |
| DeepSeek-V4-Pro | 52.7 |
| Kimi K2.7 Code | 51.4 |
| GLM-5.1 | 51.2 |
| Grok Build 0.1 | 51.0 |
| GLM-5.2 | 50.4 |
| DeepSeek-V4-Flash | 49.2 |
| Nemotron 3 Super 120B A12B | 48.4 |
| Mistral Medium 3.5 | 48.1 |
| CoBuddy | 47.1 |
| Qwen3.7-Plus | 47.1 |
| Grok 4.3 | 46.6 |
| Qwen3.6-Flash-2026-04-16 | 45.8 |
| Mistral Small 4 | 45.2 |
| Ring 2.6 1T | 44.7 |
| Gemini 3.1 Flash-Lite | 43.8 |
| Granite 4.1 8B | 43.4 |
| North Mini Code | 43.4 |
| Nemotron 3 Nano 30B A3B | 40.7 |
| Step 3.7 Flash | 38.1 |
| Nemotron 3 Nano Omni 30B A3B | 36.4 |
| MiniMax M2.7 | 32.2 |
| Gemini 3.5 Flash | 23.5 |