Run OpenAI Codex CLI on any model

OpenAI's official CLI coding agent - add your Antseed gateway as a Codex provider, or launch it with antseed codex.

Coding agentsOpenAI Responses~2 min

Codex is OpenAI's terminal coding agent. Recent versions ignore OPENAI_BASE_URL and instead read provider config from Codex settings, so Antseed plugs in as a custom model provider that speaks the Responses API.

The recommended endpoint is a gateway: add it as a provider in your user-level ~/.codex/config.toml, with the gateway key in ANTSEED_API_KEY. That works from any machine and gives Codex its own budget and usage log.

When Codex runs on the same computer as a buyer, antseed codex supplies the provider config for one run with Codex -c overrides, points it at the local buyer proxy, and leaves your real CODEX_HOME untouched.

Run Antseed first

Every integration needs an Antseed endpoint: a gateway, best on a small server so every tool on every machine can reach it, or the local buyer proxy (same computer only). One-time setup, a few minutes.

Step 1

Install OpenAI Codex CLI

Install Codex globally
npm install -g @openai/codex
Verify it runs
codex --version

Step 2

Point OpenAI Codex CLI at Antseed

Recommended: a gateway

ANTSEED_BASE_URL	https://gateway.example.com/v1
ANTSEED_API_KEY	antseed_…

Show as shell exports

export ANTSEED_BASE_URL="https://gateway.example.com/v1" export ANTSEED_API_KEY="antseed_…"

Your gateway's base URL, ending in /v1, and an API key from the Keys page of its console. Recommended: a gateway on a server, which every machine can reach. A gateway on this computer is http://localhost:8379/v1; publish it with a tunnel (antseed tunnel start) for other machines. No gateway yet: see Run Antseed first above. Export both in the shell that starts Codex.

~/.codex/config.toml

model = "@preset/coder" model_provider = "antseed" # Codex doesn't know open model ids; set the model's context length so it compacts in time. model_context_window = 128000 [model_providers.antseed] name = "Antseed gateway" base_url = "https://gateway.example.com/v1" env_key = "ANTSEED_API_KEY" wire_api = "responses"

Set base_url to your gateway URL ending in /v1; Codex reads the key from ANTSEED_API_KEY. @preset/coder is a gateway preset (create one under Presets in the console): with a raw model name, Codex can switch to a tool format some providers drop. A model id from GET /v1/models works too. Provider settings must live in your user-level config; Codex ignores them in a project-local ./.codex/config.toml.

Same computer, no gateway: the local buyer proxy

antseed codex --model deepseek-v4.1-flash

The wrapper resolves the local proxy URL, injects an Antseed model provider with wire_api = "responses", sets a placeholder ANTSEED_API_KEY, and forwards extra Codex args. Put child flags after -- when they look like wrapper flags. To route to a specific peer, prefix the model with its peer id: antseed codex --model <peerId>@deepseek-v4.1-flash.

Step 3

Pick a model

deepseek-v4.1-flashglm-5.3-flashglm-5.2kimi-k3

Set top-level model = "<model-id>" in ~/.codex/config.toml, or override per run with codex --model <model-id>. With the wrapper, pass antseed codex --model <model-id>. Route to a specific peer by prefixing its advertised service id: <peerId>@<service-id>.

The full list is every model on the network. Run GET /v1/models on your endpoint, antseed network browse, or open the live network page to see what's available right now.

Verify

Test it

List every model on the network
curl -s "$ANTSEED_BASE_URL/models" -H "Authorization: Bearer $ANTSEED_API_KEY" | jq '.data[].id'
Example response
"deepseek/deepseek-v4.1-flash" "moonshotai/kimi-k3" "z-ai/glm-5.2" "z-ai/glm-5.3-flash"
Whatever appears here, or the short name without the vendor prefix (deepseek-v4.1-flash), is a valid model value. Against the local buyer proxy: curl -s http://localhost:8377/v1/models.
Run Codex
codex
Through the local buyer proxy instead: antseed codex --model deepseek-v4.1-flash.
Verify inference is actually paid through Antseed
antseed buyer balance # local buyer proxy only
What to look for after one real prompt
Deposits available: 4.289391 USDC → 3.289391 USDC Deposits reserved: 0 USDC → 1 USDC
On a gateway, the request shows up in the console under Logs, with the seller, tokens and cost. With the local buyer proxy, a non-zero Reserved (channel opened) and/or a drop in Available (settled spend) after a real prompt confirms Antseed served the request. Do not rely on lsof -i | grep codex or ~/.codex/log/codex-tui.log: Codex keeps persistent TCP connections to Cloudflare/ChatGPT IPs for non-inference purposes, and the provider=OpenAI lines in the TUI log are not a reliable indicator that inference went to OpenAI.

How OpenAI Codex CLI talks to Antseed

Wire format sent by OpenAI Codex CLI: OpenAI Responses (hits /v1/responses on your gateway or buyer proxy)
Best-fit services: any service whose protocols array contains openai-responses. That's what the peer advertises as natively-supported - zero translation overhead, no transform edge cases.
How to check a peer: run antseed network peer <peerId> --json and look at providerServiceApiProtocols[provider].services[service]. The browse command exposes the same field per peer.
What happens when protocols don't match: Antseed's @antseed/api-adapter translates between OpenAI Responses and the service's native protocol on the fly, inside the buyer. So a request from OpenAI Codex CLI can still reach a service that only advertises anthropic-messages - just with a small transform step.

If it goes wrong

Troubleshooting

OPENAI_BASE_URL / OPENAI_API_KEY are being ignoredExpected on recent Codex builds. Add the [model_providers.antseed] block above to ~/.codex/config.toml, or use antseed codex --model <model-id> with a local buyer proxy.
401 from the gatewayExport ANTSEED_API_KEY with your gateway key in the shell that starts Codex, and keep env_key = "ANTSEED_API_KEY" in the provider block.
How can I tell if Codex is actually routing through Antseed?On a gateway, check Logs in the console after a test prompt. With the local buyer proxy, check antseed buyer balance: Reserved going from $0 to a non-zero value and/or Available dropping confirms Antseed served the request. Do not trust lsof connections to Cloudflare IPs or provider=OpenAI lines in ~/.codex/log/codex-tui.log.
Codex prints Ignored unsupported project-local config keys … model_provider, model_providersProvider settings must live in your user-level ~/.codex/config.toml. Codex rejects provider blocks in a project-local ./.codex/config.toml and falls back to its default provider.
Tool calls fail or Codex switches tool formats mid-sessionUse a gateway preset as the model (model = "@preset/coder"), as recommended in the gateway API docs.
Streaming stops after the first chunkSet wire_api = "responses" in the [model_providers.antseed] block (the wrapper does this for you).
Model metadata for … not found. Defaulting to fallback metadataExpected for open model ids. antseed codex sets a context window for you; in ~/.codex/config.toml, set model_context_window to the model's context length.
Hangs forever on first messageCheck the endpoint answers: curl -s "$ANTSEED_BASE_URL/models" -H "Authorization: Bearer $ANTSEED_API_KEY". With the local buyer proxy, also check antseed buyer status; if a stale session pin points at an offline peer, clear it with antseed buyer connection clear.

Reference

Links

Codex repo Codex sample config Antseed gateway API

Same category

Claude Code OpenCode Pi All integrations