unknown

Matt Maher tested frontier models in Cursor v. other harnesses. Cursor boosted model performance by 11% on average:

Gemini: 52% → 57%
GPT-5.4: 82% → 88%
Opus: 77% → 93%

His benchmark measures how well models implement a 100-feature PRD. @cursor_ai consistently outperformed.

<p>Matt Maher tested frontier models in Cursor v. other harnesses. Cursor boosted model performance by 11% on average:

Gemini: 52% → 57%
GPT-5.4: 82% → 88%
Opus: 77% → 93%

His benchmark measures how well models implement a 100-feature PRD. @cursor_ai consistently outperformed.</p>

jason liu

A brief, sardonic comment mocking the common practice of cramming 100 features into a single product requirements document with only a single evaluation pass — a jab at poor product planning and evaluation rigor.