Use CORe anywhere

Cursor

Cursor is an AI-first VS Code fork. Point it at CORe via Settings → Models → Override OpenAI Base URL.

1. Open Cursor Settings (Cmd/Ctrl + ,).
2. Go to Models.
3. Toggle on Override OpenAI Base URL.
4. Set the base URL to https://opencore.one/v1.
5. Set the OpenAI API Key to your ck-... key.
6. In the model list, add a custom model: core-6.2.
7. Switch the active model to core-6.2 and chat away.

Continue (VS Code / JetBrains)

Continue is an open-source coding assistant for VS Code and JetBrains. Add CORe by editing ~/.continue/config.json.

{
  "models": [
    {
      "title": "CORe 6.2",
      "provider": "openai",
      "model": "core-6.2",
      "apiBase": "https://opencore.one/v1",
      "apiKey": "ck-your-key-here",
      "contextLength": 262144
    }
  ],
  "tabAutocompleteModel": {
    "title": "CORe Warp (autocomplete)",
    "provider": "openai",
    "model": "core-warp",
    "apiBase": "https://opencore.one/v1",
    "apiKey": "ck-your-key-here"
  }
}

Save the file. Continue picks up changes automatically. Use core-warp for tab autocomplete (low latency) and core-6.2 for the chat sidebar (full reasoning).

Aider

Aider is a CLI pair-programmer that edits files directly in your repo. Use the --openai-api-base and --openai-api-key flags.

export OPENAI_API_BASE="https://opencore.one/v1"
export OPENAI_API_KEY="ck-your-key-here"

aider --model openai/core-6.2 src/

The openai/ prefix tells Aider to use the OpenAI provider; the model name after the slash is the CORe friendly id. Add --no-stream if you'd rather see whole responses at once instead of streaming.

Zed

Zed's AI assistant supports any OpenAI-compatible endpoint via its settings file.

{
  "language_models": {
    "openai": {
      "version": "1",
      "api_url": "https://opencore.one/v1",
      "available_models": [
        { "name": "core-6.2", "max_tokens": 262144 },
        { "name": "core-6.1", "max_tokens": 262144 },
        { "name": "core-warp", "max_tokens": 262144 }
      ]
    }
  },
  "assistant": {
    "default_model": { "provider": "openai", "model": "core-6.2" },
    "version": "2"
  }
}

Set the API key via Zed's Settings → AI → OpenAI API Key, or through the command palette: assistant: configure.

Open Interpreter

Open Interpreter runs code locally based on natural-language instructions. Configure with environment variables or CLI flags.

export OPENAI_API_KEY="ck-your-key-here"
export OPENAI_API_BASE="https://opencore.one/v1"

interpreter --model openai/core-6.2

For agentic workflows where you want fast iteration, swap to core-warp. For longer reasoning with thinking mode, keep core-6.2.

MSTY

MSTY is a local-first AI chat app. Add CORe under Settings → Remote Model Providers → Add Provider.

1. Open MSTY Settings.
2. Go to Remote Model Providers.
3. Click Add Provider and choose OpenAI Compatible.
4. Endpoint: https://opencore.one/v1
5. API Key: ck-...
6. Pull / refresh model list. CORe 6.2, 6.1, 4, Warp, and Turbo should appear.

cURL / Python / Node

Direct HTTP usage. The CORe API matches the OpenAI Chat Completions spec exactly, so any official OpenAI SDK works with two lines changed.

curl https://opencore.one/v1/chat/completions \
  -H "Authorization: Bearer ck-your-key-here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "core-6.2",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    base_url="https://opencore.one/v1",
    api_key="ck-your-key-here",
)

resp = client.chat.completions.create(
    model="core-6.2",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://opencore.one/v1",
  apiKey: "ck-your-key-here",
});

const resp = await client.chat.completions.create({
  model: "core-6.2",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(resp.choices[0].message.content);

Full request/response reference, tool calling, streaming, and multimodal usage are documented in the API docs.

OpenCode

OpenCode is a terminal agentic coding tool with many providers. To add us:

1. Go to the path for config (~/.config/opencode, or on Windows C:\Users\{username}\.config\opencode).
2. Create (or edit) opencode.json.
3. Paste (or append) the following:
~/.config/opencodeCopy

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "opencore": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "OpenCORe",
      "options": {
        "baseURL": "https://opencore.one/v1"
      },
      "models": {
        "xenon-9": {
          "name": "Xenon 9",
          "reasoning": true,
          "variants": {
            "low": { "reasoningEffort": "low" },
            "high": { "reasoningEffort": "high" },
            "max": { "reasoningEffort": "max" }
          }
        },
        "xenon-9s": {
          "name": "Xenon 9s",
          "reasoning": true,
          "variants": {
            "low": { "reasoningEffort": "low" },
            "high": { "reasoningEffort": "high" },
            "max": { "reasoningEffort": "max" }
          }
        },
        "xenon-9-code": {
          "name": "Xenon 9 Code",
          "reasoning": true,
          "variants": {
            "low": { "reasoningEffort": "low" },
            "high": { "reasoningEffort": "high" },
            "max": { "reasoningEffort": "max" }
          }
        },
        "xenon-8": {
          "name": "Xenon 8"
        },
        "xenon-8-code": {
          "name": "Xenon 8 Code"
        },
        "xenon-8s": {
          "name": "Xenon 8s"
        },
        "core-7-ultra": {
          "name": "CORe 7 Ultra"
        },
        "core-7-lite": {
          "name": "CORe 7 Lite"
        },
        "core-7-code": {
          "name": "CORe 7 Code"
        },
        "safertitan": {
          "name": "SaferTitan"
        },
        "agintek": {
          "name": "Agintek"
        }
      }
    }
  }
}

4. Reopen OpenCode. Use one credential source: a configured options.apiKey overrides the key saved by /connect, even when an environment variable resolves to empty. To use /connect, remove provider.opencore.options.apiKey and explicit Authorization overrides from all active global, project, and OPENCODE_CONFIG configurations. Alternatively, keep "apiKey": "{env:OPENCORE_API_KEY}" and ensure that variable contains the raw key in the process launching OpenCode.
5. Type /connect
6. In the providers, search or select OpenCORe, or choose Other and enter exactly opencore as the provider ID. Paste the full raw ck-... key, without quotes or Bearer. The shortened key shown after creation is not usable.
7. Switch the active model by using /models if you didn't already switch. Select an opencore/... model.
8. For Xenon 9, Xenon 9s, or Xenon 9 Code, use OpenCode's variant selector (default shortcut: Ctrl+T) to cycle reasoning levels. Restart OpenCode after changing its configuration.

Xenon 9 reasoning levels: low, high, and max. These are native, provider-managed effort levels, not fixed token counts. OpenCode exposes them as native variants. For apps that discover models through /v1/models, the API also lists suffixed IDs such as xenon-9-high, xenon-9s-low, and xenon-9-code-max.

Other API clients can send reasoning_effort or reasoning_budget with one of those levels. A supplied budget overrides the effort, which overrides the model suffix; base model IDs without controls leave provider defaults unchanged. The current Xenon 9 backend does not support numeric reasoning-token caps: numeric reasoning_budget or thinking_budget requests return a clear error rather than silently ignoring the cap. Reasoning levels do not change max_tokens or disable request timeouts.

Any other OpenAI-compatible tool

Most modern AI tools accept three settings: base URL, API key, and model name. If your tool has those three fields, CORe works.

• Base URL: https://opencore.one/v1
• API key field: usually labeled "OpenAI API Key" or "API token"; use your ck-...
• Model name: core-6.2 (or whichever model fits your use case)

If a tool requires a specific scheme like openai/<model> or provider/model, prefix with openai/; e.g. openai/core-6.2.