LLM2014 Code 2025-09 - Python: leaderboard
LLM2014 monthly coding benchmark for 2025-09, measuring Python code-generation problem-solving on the source score scale.
Metric: Score. Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 20 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O4 Mini (High) | 9.17 | #180 (O4 Mini) |
| 2 | GPT-5 Mini (High) | 8.96 | #186 (GPT-5 Mini) |
| 3 | Gemini 2.5 Pro | 7.5 | #152 |
| 4 | DeepSeek V3.1 (Thinking) | 7.36 | #264 (DeepSeek V3.1) |
| 5 | GLM-4.5 (Thinking) | 7.15 | #275 (GLM-4.5) |
| 6 | Claude Sonnet 4 (Thinking) | 6.94 | #199 (Claude Sonnet 4) |
| 7 | Qwen 3 235B A22B 2507 Instruct | 6.18 | #300 |
| 8 | GLM-4.5 | 6.11 | #275 |
| 9 | Qwen 3 Coder 480B A35B | 5.83 | #279 |
| 10 | Grok 4 | 5.76 | #181 |
| 11 | DeepSeek V3.1 | 5.69 | #264 |
| 12 | GPT-OSS-20B (High) | 5.62 | #518 (GPT-OSS-20B) |
| 13 | Claude Sonnet 4 | 5.42 | #199 |
| 14 | Seed-1.6 (Thinking) | 5.42 | #265 (Seed-1.6) |
| 15 | Gemini 2.5 Flash (Thinking) | 5.28 | #245 (Gemini 2.5 Flash) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm2014-code-2025-09-python · How It Works · Data refreshed daily, snapshot 2026-10-20.