> ## Documentation Index
> Fetch the complete documentation index at: https://docs.mixroute.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenWork

> OpenWork Integration Guide: Use multiple AI models in OpenWork through MixRoute

## 1. Introduction

OpenWork is an open-source AI coding assistant that supports 75+ models and local deployment. Through **MixRoute**, you can access a wide range of mainstream and cutting-edge models (such as GPT, Claude, GLM, etc.) in OpenWork, with full support for custom providers and model configuration.

**Download**: [https://openworklabs.com/](https://openworklabs.com/)

## 2. Quick Setup (MixRoute)

### 2.1 Get an API Key

Create and copy an API key from the [Mixroute API Console](https://api.mixroute.ai/).

### 2.2 Configure a Custom Provider in OpenWork

OpenWork supports adding custom providers via a JSON configuration file. It is recommended to configure at the workspace level rather than globally.

<Steps>
  <Step title="Locate the config file">
    Create or edit the config file in your OpenWork workspace:

    `/path-to-your-workspace/.config/opencode/opencode.json`

    <Warning>
      Adding custom providers to the global location `~/.config/opencode/opencode.json` is not recommended. OpenWork prioritizes workspace-level configuration; global settings will be overridden and ignored.
    </Warning>
  </Step>

  <Step title="Add a Provider">
    Add the corresponding provider configuration based on the integration method:

    <AccordionGroup>
      <Accordion title="OpenAI-Compatible Format (recommended for standard models)">
        For models supporting Chat Completions, use `@ai-sdk/openai-compatible`:

        ```json theme={null}
        {
          "$schema": "https://opencode.ai/config.json",
          "disabled_providers": [],
          "provider": {
            "mixroute-compatible": {
              "name": "mixroute-compatible",
              "npm": "@ai-sdk/openai-compatible",
              "options": {
                "baseURL": "https://api.mixroute.ai/v1",
                "apiKey": "{env:MIXROUTE_API_KEY}"
              },
              "models": {
                "glm-5.3": {
                  "name": "glm-5.3"
                }
              }
            }
          }
        }
        ```

        <Tip>
          Using `@ai-sdk/openai-compatible` gives you the broadest compatibility, covering most models that support the Chat Completions protocol.
        </Tip>
      </Accordion>

      <Accordion title="Responses API (GPT-5.x Codex and similar models)">
        Some models require the Responses API protocol. Use `@ai-sdk/openai` for these:

        ```json theme={null}
        {
          "mixroute-responses": {
            "name": "MixRoute OpenAI Responses",
            "npm": "@ai-sdk/openai",
            "options": {
              "baseURL": "https://api.mixroute.ai/v1",
              "apiKey": "{env:MIXROUTE_API_KEY}"
            },
            "models": {
              "gpt-5.6-luna": {
                "name": "gpt-5.6-luna"
              },
              "gpt-5.6-terra": {
                "name": "gpt-5.6-terra"
              },
              "gpt-5.6-sol": {
                "name": "gpt-5.6-sol"
              }
            }
          }
        }
        ```

        <Info>
          See [Section 3](#3-responses-api-required-for-some-models-important) for which models require Responses API.
        </Info>
      </Accordion>

      <Accordion title="Anthropic Messages API (Claude models)">
        Claude models connect via `@ai-sdk/anthropic`:

        ```json theme={null}
        {
          "mixroute-messages": {
            "name": "MixRoute Anthropic messages",
            "npm": "@ai-sdk/anthropic",
            "options": {
              "baseURL": "https://api.mixroute.ai/v1",
              "apiKey": "{env:MIXROUTE_API_KEY}"
            },
            "models": {
              "claude-sonnet-5": {
                "name": "claude-sonnet-5"
              }
            }
          }
        }
        ```
      </Accordion>
    </AccordionGroup>
  </Step>

  <Step title="Save and Use">
    Save the configuration file and restart OpenWork. The configured providers and models will appear in the model selection list.
  </Step>
</Steps>

**Complete configuration example** (all providers combined):

```json theme={null}
{
  "$schema": "https://opencode.ai/config.json",
  "disabled_providers": [],
  "provider": {
    "mixroute-responses": {
      "name": "MixRoute OpenAI Responses",
      "npm": "@ai-sdk/openai",
      "options": {
        "baseURL": "https://api.mixroute.ai/v1",
        "apiKey": "{env:MIXROUTE_API_KEY}"
      },
      "models": {
        "gpt-5.6-luna": {
          "name": "gpt-5.6-luna"
        },
        "gpt-5.6-terra": {
          "name": "gpt-5.6-terra"
        },
        "gpt-5.6-sol": {
          "name": "gpt-5.6-sol"
        }
      }
    },
    "mixroute-messages": {
      "name": "MixRoute Anthropic messages",
      "npm": "@ai-sdk/anthropic",
      "options": {
        "baseURL": "https://api.mixroute.ai/v1",
        "apiKey": "{env:MIXROUTE_API_KEY}"
      },
      "models": {
        "claude-sonnet-5": {
          "name": "claude-sonnet-5"
        }
      }
    },
    "mixroute-compatible": {
      "name": "mixroute-compatible",
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://api.mixroute.ai/v1",
        "apiKey": "{env:MIXROUTE_API_KEY}"
      },
      "models": {
        "glm-5.3": {
          "name": "glm-5.3"
        }
      }
    }
  }
}
```

### 2.3 Switch Models

Once configured, you can select the provider and model (e.g., `mixroute-responses/gpt-5.6-luna`) from the conversation interface or settings.

<Tip>
  * It is recommended to inject the API key via the `MIXROUTE_API_KEY` environment variable instead of writing it in plain text in the config file.
  * If using a self-hosted or alternative service, change the **base URL** to the corresponding address, e.g., `http://your-server:3003/v1`.
</Tip>

<img src="https://mintcdn.com/personal-a5418d9f/02ZDARHDo1vye0sI/images/openwork-en-1.png?fit=max&auto=format&n=02ZDARHDo1vye0sI&q=85&s=217fe7879e2430653a89623a58e93b1d" alt="Openwork En 1" width="2628" height="1766" data-path="images/openwork-en-1.png" />

### 2.4 Start OpenWork and Start a Conversation

After configuration, launch OpenWork and you will see the added models in the model selection list.

<img src="https://mintcdn.com/personal-a5418d9f/02ZDARHDo1vye0sI/images/openwork-en-2.png?fit=max&auto=format&n=02ZDARHDo1vye0sI&q=85&s=17c8cecdbc7f53bfbb23ed9309e8fc0a" alt="Openwork En 2" width="2628" height="1766" data-path="images/openwork-en-2.png" />

## 3. Responses API Required for Some Models (Important)

Some models use a different API interface from the standard Chat Completions and require the **Responses API**. If you encounter an error like this when selecting such a model in OpenWork:

> The chatCompletion operation does not work with the specified model. Please choose a different model and try again.

This means the request is being sent via **Chat Completions**, but the model only supports **Responses API** on the server side. You need to configure it to use the correct interface.

### 3.1 Models That Require Responses API (Typical)

| Model ID / Series | Description                       |
| ----------------- | --------------------------------- |
| **gpt-5.6-luna**  | GPT-5.6 Luna, Responses API only  |
| **gpt-5.6-terra** | GPT-5.6 Terra, Responses API only |
| **gpt-5.6-sol**   | GPT-5.6 Sol, Responses API only   |

### 3.2 How to Configure in OpenWork

**Add a separate provider in the config file using `@ai-sdk/openai`** to make OpenWork use the **Responses API** for these models.

See the [Responses API section in Step 2](#responses-api-gpt-5x-codex-and-similar-models) for the specific configuration. Add the `mixroute-responses` provider to your `opencode.json` file.

### 3.3 Summary

| Scenario                                                                | Approach                                                        |
| ----------------------------------------------------------------------- | --------------------------------------------------------------- |
| Using **gpt-5.6-luna / gpt-5.6-terra / gpt-5.6-sol** (Responses models) | Add a separate provider in `opencode.json` using Responses API  |
| Using **glm-5.3, claude-sonnet-5** (standard models)                    | Use `@ai-sdk/openai-compatible` or `@ai-sdk/anthropic` as usual |

## 4. Reference Links

* OpenWork Website & Download: [https://openworklabs.com/](https://openworklabs.com/)
* OpenCode Configuration & Models: [Models](https://opencode.ai/docs/models), [Providers](https://opencode.ai/docs/providers)
* MixRoute Console: [https://api.mixroute.ai/](https://api.mixroute.ai/)