unknown

Cursor is introducing a new benchmarking method for evaluating AI models on agentic coding tasks, comparing models by intelligence and efficiency metrics within the Cursor environment.

We're sharing a new method for scoring models on agentic coding tasks.

Here's how models in Cursor compare on intelligence and efficiency:

<p>We're sharing a new method for scoring models on agentic coding tasks.

Here's how models in Cursor compare on intelligence and efficiency:</p>

OpenAI Developers

GPT-5.4 has taken the top position on CursorBench, a benchmark measuring AI coding assistant performance, leading in correctness while also demonstrating efficient token usage.