Coding benchmark
AlgoTune leaderboard
As of October 2026, GPT-5.2 has the highest published AlgoTune score on Noometry at 2.05, out of 18 models with results.
Last verified
About AlgoTune
Speeding up numerical and algorithmic Python functions, scored as average speed-up over reference code.
- Category
- Coding
- Introduced
- 2025
- Format
- Code optimization
- Unit
- Raw score
- Official site
- algotune.io
Top 15 models
- GPT-5.2 2.05
- Gemini 3.1 Pro Preview 2.02
- GPT-5.4 1.85
- Gemini 3 Pro 1.83
- Claude Opus 4.5 1.77
- o4-mini 1.72
- DeepSeek-R1 1.7
- GPT-5 1.67
- Claude Sonnet 4.5 1.52
- GLM-4.5 1.52
- Gemini 2.5 Pro 1.51
- Claude Opus 4.6 1.47
- Qwen3-Coder 480B-A35B Instruct 1.44
- gpt-oss-120b 1.41
- GPT-5 Mini 1.38
Sponsored placements are available on pages like this one. Advertise on Noometry
All results
| # | Model | Provider | Rating | Setting | Source | Date |
|---|---|---|---|---|---|---|
| 1 | GPT-5.2 | OpenAI | 2.05 | medium | Epoch AI | |
| 2 | Gemini 3.1 Pro Preview |  Google | 2.02 | Epoch AI | ||
| 3 | GPT-5.4 | OpenAI | 1.85 | high | Epoch AI | |
| 4 | Gemini 3 Pro | Google | 1.83 | Epoch AI | ||
| 5 | Claude Opus 4.5 | Anthropic | 1.77 | Epoch AI | ||
| 6 | o4-mini | OpenAI | 1.72 | high | Epoch AI | |
| 7 | DeepSeek-R1 |  DeepSeek | 1.7 | Epoch AI | ||
| 8 | GPT-5 | OpenAI | 1.67 | high | Epoch AI | |
| 9 | Claude Sonnet 4.5 | Anthropic | 1.52 | Epoch AI | ||
| 10 | GLM-4.5 | Z.ai (Zhipu) | 1.52 | thinking | Epoch AI | |
| 11 | Gemini 2.5 Pro | Google | 1.51 | Epoch AI | ||
| 12 | Claude Opus 4.6 | Anthropic | 1.47 | high | Epoch AI | |
| 13 | Qwen3-Coder 480B-A35B Instruct |  Alibaba (Qwen) | 1.44 | Epoch AI | ||
| 14 | gpt-oss-120b | OpenAI | 1.41 | high | Epoch AI | |
| 15 | GPT-5 Mini | OpenAI | 1.38 | high | Epoch AI | |
| 16 | Claude Opus 4.1 | Anthropic | 1.34 | Epoch AI | ||
| 17 | Claude Opus 4 | Anthropic | 1.33 | Epoch AI | ||
| 18 | GPT-5 Pro | OpenAI | 1.31 | medium | Epoch AI |
Compare the leaders
Frequently asked questions
What does AlgoTune measure?
Speeding up numerical and algorithmic Python functions, scored as average speed-up over reference code.
Which model has the highest AlgoTune score?
As of October 2026, GPT-5.2 has the highest published AlgoTune score on Noometry at 2.05, out of 18 models with results.
What is the best open-weight model on AlgoTune?
GLM-4.5 has the highest AlgoTune score among open-weight models at 1.52, ranking 10 of 18 overall.