Hey everyone,
Big update
GPT-5.3-Codex is now live inside Pickaxe.
If youโre building anything code-heavy, this one is worth testing. GPT-5.3-Codex is designed specifically for software tasks. It handles debugging, refactoring, multi-file reasoning, and step-by-step execution much more cleanly than general models. Itโs also noticeably faster, which makes iterative testing in the Builder feel smoother.
Who should try it?
This model is especially useful for:
-
Builders creating coding assistants
-
Anyone generating or reviewing React, Python, backend APIs, or scripts
-
Builders using Actions to trigger workflows and needing precise structured outputs
-
Advanced users creating agent-style Pickaxes that plan and execute multi-step tasks
-
Technical educators building code tutors or assignment helpers
If your Pickaxe needs clean syntax, structured reasoning, or reliable code explanations, GPT-5.3-Codex is a strong option.
How to find it

-
Open your Pickaxe in the Builder
-
Go to the Prompt tab
-
Click the Model dropdown (right under the prompt box)
-
Select GPT-5.3-Codex
Thatโs it. No extra setup required.
As always, test it live in the right panel and compare outputs with your previous model. Different tools benefit from different models, so itโs worth experimenting.
Curious to hear what you all build with it. ![]()