Cursor Bench
Cursor Bench is a benchmark tool developed and maintained by the Cursor team, specifically for evaluating the coding capabilities of large language models. The benchmark systematically measures model performance in code understanding, generation, debugging, and completion through a series of realistic programming tasks and code generation scenarios, commonly used to compare the capabilities of different AI models in practical software development contexts.
Timeline (last 90 days)
On Cursor Bench, Sonnet 5.5 consumed 271,920 tokens per task, surpassing Opus's 218k and Gemini 3.8 Flash's 162k.
On Cursor Bench, Sonnet 5.5 consumed 271,920 tokens per task, surpassing Opus's approximately 218k and roughly double that of Gemini 3 Flash.
On Cursor Bench, Sonnet 5.5's best score is only about two points behind Opus 5.5
All Facts (3)
On Cursor Bench, Sonnet 5.5 consumed 271,920 tokens per task, surpassing Opus's 218k and Gemini 3.8 Flash's 162k.
50%UnverifiedOn Cursor Bench, Sonnet 5.5 consumed 271,920 tokens per task, surpassing Opus's approximately 218k and roughly double that of Gemini 3 Flash.
50%UnverifiedOn Cursor Bench, Sonnet 5.5's best score is only about two points behind Opus 5.5
50%