Codex

For Codex, register the hub as a provider in config.toml and export the token variable it names:

config.toml (in ~/.codex)

model_provider = "coderelayhub"
model = "gpt-5.6-sol"
[model_providers.coderelayhub]
name = "CodeRelay Hub"
base_url = "https://api.coderelayhub.com/v1"
env_key = "CRH_TOKEN"
wire_api="responses"

shell

export CRH_TOKEN=crh_paste_your_key_here
codex

GPT model strings to start with: gpt-5.6-sol · gpt-6-astra · gpt-6.1-sol

OpenAI SDKs

Any OpenAI-style client works once its base URL ends in /v1.

python

from openai import OpenAI as HubClient  # pip install openai

hub = HubClient(base_url="https://api.coderelayhub.com/v1", api_key="crh_paste_your_key_here")
diff_review = hub.chat.completions.create(model="gpt-6.1-sol",
    messages=[{"role": "user", "content": "Name three risks in this diff"}])
print(diff_review.choices[0].message.content)

Cursor, Cherry Studio, OpenCode and Cline / Roo take the same base URL and key in their OpenAI-compatible provider settings; Claude models stay reserved for the official Claude clients.