What is your current "Model Stack" for specific workflows?

Discussions
, , ,
1

Hey everyone,

I’m curious to see how you all are orchestrating different models for specific phases of development within Cursor. We have so many specialized options now, and “one model to rule them all” doesn’t seem efficient anymore.

Here is my current setup:

  • Planning: Claude Opus 4.5 Thinking

  • Implementation: Claude Opus 4.5

  • Quick Fixes & Queries: Composer 1

  • Documentation & Text Gen: Gemini 3 Pro

I’ve been eyeing ChatGPT 5.2 Codex. The benchmarks and snippets look impressive for pure code generation, but I’m hesitant to break my muscle memory and established workflow.

What are you guys running currently?

3

Planning: GPT-5.2-High

Executing: GPT-5.2-Codex-High

Questions/Quick Fixes: Composer 1

Small/medium tasks/UI: Opus 4.5

2 Likes
4

Anything “tricky” I use Claude Opus. I probably wouldn’t notice the difference if I was using Codex

My main workhorse is Composer-1 for frontend work and backend plumbing work. I’d say about 90% I use this model but its not cheap.

”Auto” towards the end of the month or any donkey work I don’t mind waiting for

1 Like