BLXBench Reasoning category
Code · 2026-04-25
Reasoning category component of the BLXBench public leaderboard. The source describes this category as arithmetic, symbolic steps, and structured problem-solving; scores are category score percentages.
Top models (higher is better)
| Model | Score |
|---|---|
| Opus 4.8 | 79.5 |
| GPT-5.5 | 76.2 |
| Opus 4.7 | 75.6 |
| Claude Fable 5 | 75.4 |
| GPT-5.3-Codex | 74.7 |
| Kimi K2.7 Code | 73.6 |
| Granite 4.1 8B | 73.4 |
| DeepSeek-V4-Pro | 73.2 |
| MiMo-V2.5 | 73.2 |
| Grok Build 0.1 | 72.3 |
| MiniMax M3 | 71.7 |
| MiMo-V2.5-Pro | 71.5 |
| Qwen3.6-Flash-2026-04-16 | 70.4 |
| Qwen3.7-Max | 69.9 |
| Ring 2.6 1T | 69.6 |
| CoBuddy | 69.3 |
| Mistral Small 4 | 69.2 |
| DeepSeek-V4-Flash | 69.1 |
| Qwen3.7-Plus | 68.0 |
| Kimi K2.6 | 67.8 |
| Mistral Medium 3.5 | 67.8 |
| Nemotron 3 Nano 30B A3B | 67.5 |
| Step 3.7 Flash | 67.1 |
| Nemotron 3 Super 120B A12B | 66.5 |
| North Mini Code | 66.5 |
| Grok 4.3 | 63.6 |
| GLM-5.2 | 63.2 |
| GLM-5.1 | 60.4 |
| MiniMax M2.7 | 59.5 |
| Nemotron 3 Nano Omni 30B A3B | 51.9 |
| Gemini 3.5 Flash | 50.5 |
| Gemini 3.1 Flash-Lite | 49.9 |