gpt-6POST
Developer uses cheap models for daily coding, saves Astra for design-partner work
Matteo Collina says 80% of his code comes from OpenAI Luna or DeepSeek Flash, and that using Astra, Sol or Fable on routine churn is a waste, reserving the best model for when he needs a design partner.
Matteo Collina
@matteocollinaArchived 2026.09.20Originally posted 2026.09.19
Matteo Collina @matteocollinaOriginally posted 2026.09.19
80% of the code I produce is done via OpenAI Luna or DeepSeek Flash. 80% of the value I generate is done via Fable or Sol (sorry, still making sense of Astra). I feel that using Astra, Sol, Fable on most of my churn is total waste. My rule of thumb: if I know what I’m doing, the fast&cheap models are better choice. If I need a design partner, I go for the best I can afford.
This post quotes another post: x.com/zeeg/status/2098064539170623954
x.com/matteocollina/status/2101289837261189134Archived 2026.09.20 10:08