CoreCodeDocs

Connect

OpenAI-compatible API

Last updated: October 28, 2026

Every CoreCode model is available in the OpenAI API format: the natural fit for GPT, Gemini and Grok, and it works for Claude too. The OpenAI SDK, curl, and any OpenAI-compatible client that lets you enter a base URL and a key can connect to CoreCode directly.

Base URL
https://corecode.live/v1
Protocol
Chat Completions (/v1/chat/completions) and Responses (/v1/responses), for every model
Key
Header Authorization: Bearer sk-your-key, in the model's group or in auto
Models
Listed per group below; the Pricing page is the authority

Create a key

On the API Keys page, click "Create API key" and choose the group for the models you want:

ModelsGroup
GPTGPT
GeminiGemini
GrokGrok
Several of themauto, which picks the group from the model name

Try it with curl

curl https://corecode.live/v1/chat/completions \
  -H "Authorization: Bearer sk-your-key" \
  -H "Content-Type: application/json" \
  -d '{"model": "grok-4.7", "messages": [{"role": "user", "content": "reply ok"}]}'

If choices[0].message.content is ok, you are connected. Change model to gpt-6.1-sol or gemini-3.8-flash-high to try another group; the key's group has to match.

OpenAI SDK

from openai import OpenAI

client = OpenAI(base_url="https://corecode.live/v1", api_key="sk-your-key")
resp = client.chat.completions.create(
    model="gemini-3.8-flash-high",
    messages=[{"role": "user", "content": "reply ok"}],
)
print(resp.choices[0].message.content)

The Responses API works too, for every model: client.responses.create(model="gpt-6.1-sol", input="reply ok").

In a client app

Chat apps, editor plugins and agent tools name things differently. Find a provider of type "OpenAI" or "OpenAI-compatible" and fill in three fields:

FieldValue
API address / Base URLhttps://corecode.live/v1. Some clients append /v1 themselves and want https://corecode.live; if requests return 404, try the other form
API keyYour key, starting with sk-
ModelType the model name, for example gpt-6.1-sol, gemini-3.8-flash-high, grok-4.7

Tool-specific setups are in OpenCode, OpenClaw and Codex; with Magpie installed, picking models there is simpler.

GPT group

A Codex pool. Models: gpt-6.1-sol, gpt-6.1-terra, gpt-6-astra, gpt-6-sol, gpt-6-luna, gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna, codex-auto-review.

Gemini group

An Antigravity pool. Models: gemini-3.8-flash-high.

Grok group

A Grok pool. Models: grok-4.7, grok-4.6.

FAQ

401

The key is missing or wrong. The header is Authorization: Bearer sk-your-key, with one space after Bearer.

Model not found or no available channel

The key's group does not serve that model. A Grok key calling gpt-6.1-sol, for example, gets this error; use a key in the right group, or one in auto.

Can I call Claude through this API?

Yes, with a key in 「Claude 满血(可外接)」 or auto; 仅限CC admits Claude Code only. Claude's prompt caching and thinking are most complete on the native Anthropic API, though, so from an SDK prefer the Anthropic SDK at https://corecode.live.