Noometry

Coding benchmark

CursorBench leaderboard

As of October 2026, Claude Opus 5.5 has the highest published CursorBench score on Noometry at 57.8%, out of 14 models with results.

Last verified

About CursorBench

Coding-agent tasks drawn from real Cursor sessions, published by Cursor.

Category
Coding
Introduced
2026
Unit
Percent (random guessing ≈ 0%)
Official site
cursor.com

Top 14 models

  1. Claude Opus 5.5 57.8%
  2. Claude Sonnet 5.5 55.5%
  3. Claude Fable 5.1 51.8%
  4. Claude Opus 5 46.6%
  5. Grok 4.7 46.3%
  6. GLM-5.3 42.6%
  7. GPT-5.6 Sol 41.7%
  8. Muse Spark 1.3 41.6%
  9. Grok 4.6 41.4%
  10. GPT-5.6 Terra 41.3%
  11. Gemini 3.8 Flash 39.6%
  12. GLM-5.3-Flash 36.8%
  13. GPT-5.6 Luna 35.9%
  14. Claude Sonnet 5 34.1%

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

All results

#ModelProviderScoreSettingSourceDate
1Claude Opus 5.5 Anthropic57.8%maxEpoch AI
2Claude Sonnet 5.5 Anthropic55.5%maxEpoch AI
3Claude Fable 5.1 Anthropic51.8%maxEpoch AI
4Claude Opus 5 Anthropic46.6%maxEpoch AI
5Grok 4.7 xAI46.3%xhighEpoch AI
6GLM-5.3 Z.ai (Zhipu)42.6%maxEpoch AI
7GPT-5.6 Sol OpenAI41.7%maxEpoch AI
8Muse Spark 1.3 Meta41.6%maxEpoch AI
9Grok 4.6 xAI41.4%xhighEpoch AI
10GPT-5.6 Terra OpenAI41.3%maxEpoch AI
11Gemini 3.8 Flash Google39.6%highEpoch AI
12GLM-5.3-Flash Z.ai (Zhipu)36.8%maxEpoch AI
13GPT-5.6 Luna OpenAI35.9%maxEpoch AI
14Claude Sonnet 5 Anthropic34.1%maxEpoch AI

Compare the leaders

Other coding benchmarks

Frequently asked questions

What does CursorBench measure?

Coding-agent tasks drawn from real Cursor sessions, published by Cursor.

Which model has the highest CursorBench score?

As of October 2026, Claude Opus 5.5 has the highest published CursorBench score on Noometry at 57.8%, out of 14 models with results.

What is the best open-weight model on CursorBench?

GLM-5.3 has the highest CursorBench accuracy among open-weight models at 42.6%, ranking 6 of 14 overall.