Noometry

Coding benchmark

DeepSWE leaderboard

As of October 2026, GPT-6.1 Sol has the highest published DeepSWE score on Noometry at 75.2%, out of 29 models with results.

Last verified

About DeepSWE

A description with primary sources is being prepared for this benchmark.

Category
Coding
Introduced
2026
Format
Repository patch
Unit
Percent (random guessing ≈ 0%)
Official site
epoch.ai

Top 15 models

  1. GPT-6.1 Sol 75.2%
  2. GPT-6 Astra 74.1%
  3. Gemini 3.8 Flash 73.8%
  4. Claude Opus 5 73.6%
  5. GPT-5.6 Sol 72.7%
  6. Claude Fable 5 69.9%
  7. GPT-5.6 Terra 69.6%
  8. GLM-5.3 69%
  9. GPT-6 Sol 68.8%
  10. Kimi K3 68.5%
  11. Grok 4.6 67.5%
  12. GPT-5.6 Luna 67.2%
  13. GPT-5.5 67%
  14. GPT-6 Luna 66.6%
  15. Gemini 3.7 Flash 65.5%

Sponsored placements are available on pages like this one. Advertise on Noometry

All results

#ModelProviderScoreSettingSourceDate
1GPT-6.1 Sol OpenAI75.2%highModel card (self-reported)2026-09-29
2GPT-6 Astra OpenAI74.1%xhighEpoch AI
3Gemini 3.8 Flash Google73.8%highEpoch AI
4Claude Opus 5 Anthropic73.6%maxEpoch AI
5GPT-5.6 Sol OpenAI72.7%maxEpoch AI
6Claude Fable 5 Anthropic69.9%xhighEpoch AI
7GPT-5.6 Terra OpenAI69.6%maxEpoch AI
8GLM-5.3 Z.ai (Zhipu)69%maxEpoch AI
9GPT-6 Sol OpenAI68.8%maxModel card (self-reported)2026-09-22
10Kimi K3 Moonshot AI68.5%maxEpoch AI
11Grok 4.6 xAI67.5%mediumEpoch AI
12GPT-5.6 Luna OpenAI67.2%maxEpoch AI
13GPT-5.5 OpenAI67%xhighEpoch AI
14GPT-6 Luna OpenAI66.6%maxModel card (self-reported)2026-09-22
15Gemini 3.7 Flash Google65.5%mediumEpoch AI
16GLM-5.3-Flash Z.ai (Zhipu)63.4%maxEpoch AI
17Claude Opus 4.8 Anthropic59%maxEpoch AI
18Qwen3.8 Max Alibaba (Qwen)57.5%xhighEpoch AI
19Muse Spark 1.2 Meta54.9%xhighEpoch AI
20Claude Sonnet 5 Anthropic53.8%maxEpoch AI
21Grok 4.5 xAI53.8%highEpoch AI
22Muse Spark 1.1 Meta53.3%Epoch AI
23GPT-5.4 OpenAI51.8%xhighEpoch AI
24Gemini 3.6 Flash Google46.7%highEpoch AI
25GLM-5.2 Z.ai (Zhipu)43.8%maxEpoch AI
26Gemini 3.5 Flash Google37.4%mediumEpoch AI
27Kimi K2.7 Code Moonshot AI30.5%Epoch AI
28Claude Sonnet 4.6 Anthropic29.9%highEpoch AI
29Gemini 3.1 Pro Preview Google11.7%Epoch AI

Compare the leaders

Other coding benchmarks

Frequently asked questions

Which model has the highest DeepSWE score?

As of October 2026, GPT-6.1 Sol has the highest published DeepSWE score on Noometry at 75.2%, out of 29 models with results.

What is the best open-weight model on DeepSWE?

GLM-5.3 has the highest DeepSWE accuracy among open-weight models at 69%, ranking 8 of 29 overall.