Using several AI models to plan and review code in Cursor
A is looking at Cursor as a way to run several AI models on the same development work. The idea is to use Cursor CLI as a harness for models such as Opus, GPT, Flash, or Composer.
The same has already worked well in Claude Code and Codex, with better perceived quality. The main tasks are to check and improve a plan for feasibility, correctness, and , then check and improve the for correctness, simplicity, and .
A Reddit Fusion benchmark is mentioned as a possible signal that combining Opus and GPT scored higher than another model called Fable. The tradeoff is that this workflow is slow, somewhat tedious, and expensive.
Key points
- Cursor CLI is being considered as a harness for multi-model coding work.
- The workflow compares Cursor with Claude Code and Codex experience.
- The main review targets are feasibility, correctness, , and simplicity.
- The cited example suggests Opus plus GPT may perform better together in a benchmark.
- The downside is slower work, more manual effort, and higher cost.