CursorBench 4.0
A benchmark introduced by Cursor for evaluating model performance in real coding workflows.
Timeline (last 90 days)
Claude Fable 5.2 Max scored 60% on CursorBench 4.0, second only to carbon high's 60.4%.
The score gap between carbon high and Fable Max on CursorBench 4.0 is only 0.4 percentage points.
Gemini 4 Carbon ranks first on the CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench benchmarks, and its cost per task is more than 2 times lower than the next-best external model.
On the fifth edition of the Comprehensive Intelligence leaderboard, Gemini 4 Carbon scored 59 points, ranking first and overtaking Claude Opus 5.5, 9 points higher than Argon.
Cursor has launched CursorBench 4.0 for evaluating agent performance on real-world, ambiguous, multi-file tasks, and the addition of long-horizon problems has caused scores to drop across all models.
All Facts (4)
Claude Fable 5.2 Max scored 60% on CursorBench 4.0, second only to carbon high's 60.4%.
50%UnverifiedThe score gap between carbon high and Fable Max on CursorBench 4.0 is only 0.4 percentage points.
50%UnverifiedGemini 4 Carbon ranks first on the CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench benchmarks, and its cost per task is more than 2 times lower than the next-best external model.
50%UnverifiedCursor has launched CursorBench 4.0 for evaluating agent performance on real-world, ambiguous, multi-file tasks, and the addition of long-horizon problems has caused scores to drop across all models.
50%