Skip to content
forgedocs

Using OpenAI Codex and Other Models

Configure and optimize OpenAI Codex inline models and terminal assistants using the Forge runtime.

Last updated: 16 Aug 2026Runtime Wrappers

Codex and Inline Co-pilots

Inline co-pilots require low-latency completions. If the context payload is bloated, code completions take seconds to load. Forge helps by keeping the active window footprint to a minimum.

Launching Codex Wrappers

To run Forge with Codex or default OpenAI completion endpoints, execute:

forge codex

Performance Benefits

By pruning unnecessary headers and imports, Forge decreases API request payloads. This results in:

  • Lower network transmission times.
  • Reduced inference pricing model costs.
  • Sub-second code completion responses.