Codex Custom Provider: model_provider, Base URL, and wire_api

Short answer

model_provider tells Codex CLI which provider to use. Its value must match a provider id defined under [model_providers.<id>]. base_url is the API endpoint where model requests are sent. For an OpenAI-compatible provider, it usually points to a /v1 style endpoint. wire_api tells Codex which protocol shape to use when talking to that provider. In the Codex configuration reference, responses is a supported provider protocol value.

When this matters

You are configuring a custom OpenAI-compatible provider
Codex says the provider cannot be found or model requests fail
You are not sure how model_provider differs from model
You are putting BetterToken, OpenAI, or another provider in one config.toml
You need to separate Base URL, API Key, and Model ID

Field relationships

Field	Purpose
`model_provider`	Selects the active provider id
`[model_providers.<id>]`	Defines one provider’s detailed configuration
`name`	Display name for the provider
`base_url`	Root API endpoint
`wire_api`	Provider protocol, commonly `responses`
`model`	Default main model ID
`review_model`	Model used for code review flows such as `/review`

Recommended pattern

Keep the provider id consistent:

model_provider = "custom"
model = "gpt-5.5"

[model_providers.custom]
name = "BetterToken"
base_url = "https://bettertoken.ai/v1"
env_key = "MODEL_PROVIDER_API_KEY"
wire_api = "responses"
requires_openai_auth = false

This means Codex uses the provider named custom, sends model requests to https://bettertoken.ai/v1, and communicates through the Responses API shape. Do not put the API Key directly in config.toml. Set it in the terminal that launches Codex:

export MODEL_PROVIDER_API_KEY="YOUR_API_KEY"

On Windows PowerShell:

$env:MODEL_PROVIDER_API_KEY="YOUR_API_KEY"

Verify the setup

Confirm that model_provider = "custom" exactly matches [model_providers.custom].
Confirm that the Base URL is https://bettertoken.ai/v1, including /v1.
Copy a current Model ID from the GPT provider in the model plaza.
Fully exit Codex and relaunch it from a new terminal where the environment variable is set.
Send a small code-explanation task and confirm that Codex returns a complete response.

Troubleshooting

Symptom	Cause	Fix
Provider not found	The provider id and configuration table do not match	Use `custom` in both places
Missing API Key at startup	The variable named by `env_key` is unavailable	Set `MODEL_PROVIDER_API_KEY`, then open a new terminal
`401` or `403`	Invalid API Key or mixed OpenAI authentication	Copy the Key again and set `requires_openai_auth = false`
`404`	The Base URL lacks `/v1` or points to an Anthropic endpoint	Use `https://bettertoken.ai/v1`
Model not found	The value is a display name, retired model, or non-GPT model	Copy a current Model ID from the model plaza
Responses protocol error	`wire_api` is missing or incorrect	Set `wire_api = "responses"`

Claude Code

Claude Desktop

Codex

Cursor

Cline

OpenCode

Model calls

Service assurance

Cost and tokens

Concepts

Codex Custom Provider: model_provider, Base URL, and wire_api

Short answer

When this matters

Field relationships

Recommended pattern

Verify the setup

Troubleshooting

References

​Short answer

​When this matters

​Field relationships

​Recommended pattern

​Verify the setup

​Troubleshooting

​Related docs

​References

Short answer

When this matters

Field relationships

Recommended pattern

Verify the setup

Troubleshooting

Related docs

References