Benchmarks

AI coding benchmark analysis and leaderboards, comparing models on correctness and cost to help you pick a model stack.

Published on
June 25, 2026
AI Coding Benchmark Leaderboard: Cursor + DeepSWE
A combined CursorBench and DeepSWE leaderboard comparing Opus 5, GPT-5.6, Kimi K3, Fable 5, and other coding models on correctness and cost.
ai llm benchmarks coding-agents cursor