# SWE-bench Verified leaderboard: AI models ranked

- **URL canônica:** https://meirinho.com.br/en/ia/modelos/benchmark/swe-bench-verified/
- **Idioma:** en
- **Atualizado:** 2026-11-17

---
AI models

# SWE-bench Verified leaderboard: AI models ranked

31 models ranked. The top score on SWE-bench Verified belongs to Claude Opus 4.7, at 83.5% per Epoch AI, and 4 models are in a statistical tie with it.

SWE-bench Verified, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is 500 real issues from Python repositories on GitHub. The model has to produce the fix and pass the automated tests. The score is the share of tasks solved. Here are the 31 models in this section with a score on SWE-bench Verified, per Epoch AI. Rank is shown as a range, because scores within the margin are in a statistical tie.

SWE-bench Verified

83.5%

Claude Opus 4.7

models

31

| Model | % | Source |
| --- | --- | --- |
| [Claude Opus 4.7](https://meirinho.com.br/en/ia/modelos/claude-opus-4-7/)Anthropic · closed | **83.5%** 1st of 31 top score | Epoch AI |
| [GPT-5.5](https://meirinho.com.br/en/ia/modelos/gpt-5-5/)OpenAI · closed | **80.6%** 2nd of 31 | Epoch AI |
| [Gemini 3.5 Flash](https://meirinho.com.br/en/ia/modelos/gemini-3-5-flash/)Google · closed | **79.3%** between 3rd and 5th of 31 | Epoch AI |
| [Claude Opus 4.6](https://meirinho.com.br/en/ia/modelos/claude-opus-4-6/)Anthropic · closed | **78.7%** between 3rd and 5th of 31 | Epoch AI |
| [GLM-5.2](https://meirinho.com.br/en/ia/modelos/glm-5-2/)Zhipu AI (Z.ai) · open | **78.7%** between 3rd and 5th of 31 | Epoch AI |
| [DeepSeek V4 Pro](https://meirinho.com.br/en/ia/modelos/deepseek-v4-pro/)DeepSeek · open | **77.6%** between 6th and 11th of 31 | Epoch AI |
| [Qwen3.7 Max](https://meirinho.com.br/en/ia/modelos/qwen3-7-max/)Alibaba · closed | **77.3%** between 6th and 11th of 31 | Epoch AI |
| [GPT-5.4](https://meirinho.com.br/en/ia/modelos/gpt-5-4/)OpenAI · closed | **76.9%** between 6th and 11th of 31 | Epoch AI |
| [Claude Opus 4.5](https://meirinho.com.br/en/ia/modelos/claude-opus-4-5/)Anthropic · closed | **76.7%** between 6th and 11th of 31 | Epoch AI |
| [Kimi K2.6](https://meirinho.com.br/en/ia/modelos/kimi-k2-6/)Moonshot AI · open | **76.7%** between 6th and 11th of 31 | Epoch AI |
| [Qwen 3.6 Max (Preview)](https://meirinho.com.br/en/ia/modelos/qwen-3-6-max-preview/)Alibaba · undisclosed | **76.7%** between 6th and 11th of 31 | Epoch AI |
| [Gemini 3.1 Pro](https://meirinho.com.br/en/ia/modelos/gemini-3-1-pro/)Google · closed | **75.6%** between 12th and 15th of 31 | Epoch AI |
| [Gemini 3 Flash](https://meirinho.com.br/en/ia/modelos/gemini-3-flash/)Google · closed | **75.4%** between 12th and 15th of 31 | Epoch AI |
| [Claude Sonnet 4.6](https://meirinho.com.br/en/ia/modelos/claude-sonnet-4-6/)Anthropic · closed | **75.2%** between 12th and 16th of 31 | Epoch AI |
| [GPT-5.3-Codex](https://meirinho.com.br/en/ia/modelos/gpt-5-3-codex/)OpenAI · closed | **74.8%** between 12th and 18th of 31 | Epoch AI |
| [GLM 5.1](https://meirinho.com.br/en/ia/modelos/glm-5-1/)Zhipu AI (Z.ai) · open | **74.2%** between 14th and 20th of 31 | Epoch AI |
| [GPT-5.2](https://meirinho.com.br/en/ia/modelos/gpt-5-2/)OpenAI · closed | **73.8%** between 15th and 21st of 31 | Epoch AI |
| [Kimi K2.5](https://meirinho.com.br/en/ia/modelos/kimi-k2-5/)Moonshot AI · open | **73.8%** between 15th and 21st of 31 | Epoch AI |
| [GPT-5](https://meirinho.com.br/en/ia/modelos/gpt-5/)OpenAI · closed | **73.6%** between 16th and 21st of 31 | Epoch AI |
| [Claude Opus 4.1](https://meirinho.com.br/en/ia/modelos/claude-opus-4-1/)Anthropic · closed | **73.3%** between 16th and 21st of 31 | Epoch AI |
| [Gemini 3 Pro](https://meirinho.com.br/en/ia/modelos/gemini-3-pro/)Google · closed | **72.9%** between 17th and 22nd of 31 | Epoch AI |
| [GLM 5](https://meirinho.com.br/en/ia/modelos/glm-5/)Zhipu AI (Z.ai) · open | **72.1%** between 21st and 23rd of 31 | Epoch AI |
| [Claude Sonnet 4.5](https://meirinho.com.br/en/ia/modelos/claude-sonnet-4-5/)Anthropic · closed | **71.3%** between 22nd and 24th of 31 | Epoch AI |
| [Claude Opus 4](https://meirinho.com.br/en/ia/modelos/claude-opus-4/)Anthropic · closed | **70.7%** between 23rd and 24th of 31 | Epoch AI |
| [GPT-5.1](https://meirinho.com.br/en/ia/modelos/gpt-5-1/)OpenAI · closed | **68%** 25th of 31 | Epoch AI |
| [GPT-5 Mini](https://meirinho.com.br/en/ia/modelos/gpt-5-mini/)OpenAI · closed | **64.7%** 26th of 31 | Epoch AI |
| [o3](https://meirinho.com.br/en/ia/modelos/o3/)OpenAI · closed | **62.3%** 27th of 31 | Epoch AI |
| [Claude 3.7 Sonnet](https://meirinho.com.br/en/ia/modelos/claude-3-7-sonnet/)Anthropic · closed | **61%** 28th of 31 | Epoch AI |
| [Qwen3.6 Plus](https://meirinho.com.br/en/ia/modelos/qwen3-6-plus/)Alibaba · closed | **57.9%** between 29th and 30th of 31 | Epoch AI |
| [Gemini 2.5 Pro](https://meirinho.com.br/en/ia/modelos/gemini-2-5-pro/)Google · closed | **57.6%** between 29th and 30th of 31 | Epoch AI |
| [GPT-4.1](https://meirinho.com.br/en/ia/modelos/gpt-4-1/)OpenAI · closed | **48.5%** 31st of 31 | Epoch AI |

## FAQ

### Which model has the top score on SWE-bench Verified?

The top score on SWE-bench Verified belongs to Claude Opus 4.7, at 83.5% per Epoch AI, and 4 models are in a statistical tie with it.

### What does SWE-bench Verified measure?

SWE-bench Verified, on the Epoch AI leaderboard (CC BY 4.0, latest version we checked), is 500 real issues from Python repositories on GitHub. The model has to produce the fix and pass the automated tests. The score is the share of tasks solved.

### What does “statistical tie with the top score” mean?

It means the gap to the top score is within the 95% margin, which is never below 1 point. On SWE-bench Verified, near 90%, that margin is about 3.8 points. That is why rank is shown as a range, «between Xth and Yth of N»: among tied models, the order may be measurement noise.

### Which models are in this ranking?

The 31 models in this section with a published score on SWE-bench Verified, per Epoch AI. No version inherits another's score.

## How to read

1. Benchmark scores are percentages, and higher is better.
2. Rank is shown as a range, «between 1st and 5th of 31». Within the range, the gap to neighboring models falls within the benchmark's margin. The «top score» badge marks the model with the benchmark's top score, and the «statistical tie with the top score» badge marks models whose gap to it is within the margin.
3. On a saturated benchmark, the table shows the scores with the badge and without a rank order.

## Sources

- [Epoch AI](https://epoch.ai/benchmarks), CC BY 4.0, latest version we checked.
- [OpenRouter Benchmarks](https://openrouter.ai/benchmarks), CC BY 4.0, latest version we checked.
- [OpenRouter](https://openrouter.ai/api/v1/models), dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.

Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.

## How to cite

If you use this data, cite the page and the date you accessed it.

```
Adriano Meirinho. "SWE-bench Verified leaderboard: AI models ranked". meirinho.com.br. https://meirinho.com.br/en/ia/modelos/benchmark/swe-bench-verified/. Accessed November 21, 2026.
```

Third-party data follows the license of each source. The full table is available as [CSV](https://meirinho.com.br/ia/modelos/dados.csv) and [JSON](https://meirinho.com.br/ia/modelos/dados.json).

[#](#)
