Models
Step 1 of 2. Choose how many coding agents you will run and the model that serves them. Hardware and launch come next.
Model
53 curated · any Hugging Face repoPrefer to do this from your terminal? Set up and run from Codex.
How many coding agents will you run at once?
Each build below is a measured launch; the speed shown is what one agent gets when that many run together.
All measured builds measured on real hardware · all runs
More on Hugging Face
Answers 200 when ready; every other path needs the endpoint key, checked at the edge.
Registry credentials are used once for the pull and never stored.
—
Weights as stored—
Parameters—
License—
Create a read token at huggingface.co. Used once to check access, kept for the build for 30 days, never logged.
Model wizard
Describe the job. It ranks 53 models against the hardware you can use and shows measured numbers where we have them.
ranked by the Inferoptimal serving estimator · measured rows from console results