Codex NIM Proxy
A single-file, dependency-free transparent proxy that lets the Codex CLI or desktop app call NVIDIA NIM's free-hosted models, such as DeepSeek V4 Pro and Qwen3 Coder, using a user's own NVIDIA account.
Codex NIM Proxy sits between Codex, OpenAI's coding assistant, and NVIDIA NIM, converting requests and streamed responses between Codex's Responses API and NVIDIA's Chat Completions API so Codex can call models NVIDIA hosts for free, including DeepSeek V4 Pro, Qwen3 Coder, Kimi K2.6 and several others. It is a single Node.js file with no external dependencies, listening locally and automatically restoring Codex's original configuration when the proxy is closed.
A local web UI lets a user fetch the current list of available NIM models in real time, switch the active model with one click, and see the request flow between Codex and NVIDIA in a sequence diagram. It also includes a built-in DuckDuckGo web-search tool for Codex, automatic retry on transient failures, and an automatically maintained blacklist of models that return errors.