Skip to main content
FireConnect routes the OpenAI Codex CLI through Fireworks AI models via the Responses API. See the FireConnect overview for install and CLI basics.

​
Prerequisites

Fire Pass keys (fpk_...) are not supported for Codex yet. The /responses endpoint requires a standard Fireworks API key.

​
Enable Fireworks routing

export FIREWORKS_API_KEY=fw_...
fireconnect codex on
Or pass the key once so later model commands reuse it: 
fireconnect codex on --api-key fw_...
After fireconnect codex on, off, model select, or model reset, config.toml is updated immediately. To use the updated routing in Codex, run /model in the same session, start a new session, or /exit and resume with codex resume <id>.

​
Default model

Codex routes a single default model. The default is glm-latest. 
fireconnect codex status

​
What gets written

FireConnect edits ~/.codex/config.toml:
  • Sets root model_provider / model for Codex 0.134+
  • Adds a [model_providers.fireworks-ai] block with wire_api = "responses"
  • With --api-key or a key from ~/.fireconnect/config.json, writes experimental_bearer_token so later model list, select, and reset work without passing the key again
  • With only FIREWORKS_API_KEY in the environment, writes env_key = "FIREWORKS_API_KEY" instead
FireConnect snapshots your original ~/.codex/config.toml before the first change. The snapshot lives in ~/.fireconnect/codex/. Running fireconnect codex off restores it byte-for-byte. Unrelated Codex settings (for example [[mcp_servers]]) are preserved via surgical TOML edits.

​
Codex model catalog

When you run fireconnect codex on, FireConnect fetches your account’s serverless catalog and writes Codex-compatible model metadata to ~/.codex/fireworks-model-catalog.json. It links that file from config.toml via model_catalog_json so Codex knows display names, context windows, reasoning levels, and tool-calling support for each model. 
fireconnect codex status   # shows whether the catalog is linked and on disk
The catalog includes serverless models that support tool calling with a non-zero context window, plus curated aliases such as glm-latest. Deprecated models are excluded. If catalog generation fails (for example, due to an invalid API key), routing still works but Codex may show limited model metadata until you re-run fireconnect codex on. fireconnect codex model list and model select filter the picker to models present in the catalog. JSON output from model list includes Fireworks IN / OUT pricing where known.

​
Browsing and picking models

fireconnect codex model list              # browse Codex-compatible serverless endpoints
fireconnect codex model select            # pick Codex's default model
fireconnect codex model select --search glm
fireconnect codex model list --json       # includes pricing metadata where known

​
CLI reference

fireconnect codex on              # Enable Fireworks routing
fireconnect codex off             # Restore original config
fireconnect codex status          # Check current provider and model
fireconnect codex model list      # Browse serverless endpoints
fireconnect codex model select    # Pick a model interactively
fireconnect codex model reset     # Reset model to default
fireconnect codex help            # Show harness-specific help
Run fireconnect codex help for all options.

​
Switch models

fireconnect codex on --main glm-5p2
fireconnect codex on --main deepseek-v4-flash
Some models expose multiple reasoning levels in the Codex catalog (for example, glm-5p2 supports high and max). Pick the model in Codex with /model after switching.

​
Turn off Fireworks routing

fireconnect codex off
This restores your previous config.toml from the backup in ~/.fireconnect/codex/.

​
Use a non-default config file

fireconnect codex on --config-path /path/to/config.toml

​
Fireworks on Microsoft Foundry

Codex supports Fireworks on Microsoft Foundry (CLI: --provider azure or on --azure). See the FireConnect overview and Microsoft Foundry integration guide for portal setup.
Foundry routing requires a standard Azure API key. Fire Pass keys (fpk_...) are not supported. FireConnect does not write a Fireworks model catalog on the Foundry path — set your deployment with --main.

​
Configure and enable

export AZURE_API_KEY=<your-azure-api-key>

fireconnect configure \
  --provider azure \
  --base-url https://<resource>.services.ai.azure.com \
  --api-key $AZURE_API_KEY \
  --harnesses codex

fireconnect codex on --main FW-GLM-5.1
One-off routing: 
fireconnect codex on --azure --base-url https://<resource>.services.ai.azure.com --main FW-MiniMax-M2.5

​
What gets written

FireConnect sets model_provider = "fireworks-azure" in config.toml with a Fireworks on Microsoft Foundry provider block pointed at your Foundry endpoint. The Azure API key is stored as env_key = "AZURE_API_KEY" when resolved from the environment, or written literally when passed with --api-key. Pass your Foundry deployment name with --main. model list and model select browse the Fireworks serverless catalog only.

​
Source

FireConnect is open source: github.com/fw-ai/fireconnect