Use VS Code Copilot

Overview

This guide explains how to add RoutinAI custom models to VS Code Copilot and verify that the models work correctly. After configuration, you can select models whose names start with RoutinAI directly in Copilot Chat.

Prerequisites

Visual Studio Code is installed.
GitHub Copilot / Copilot Chat is installed and signed in.
You have a valid RoutinAI API key.

Configure RoutinAI Models

1. Open Copilot settings

Open Copilot Chat in VS Code, then click the settings entry in the chat panel, as shown below.

2. Open the model configuration file

Click the file icon in the upper-right corner to open the Copilot model configuration file. Use the image below as a reference.

3. Add the model configuration

Replace or merge the configuration file content with the JSON below. Keep {ApiKey} in the URL; Copilot will use your configured key value when sending requests.

[
  {
    "name": "RoutinAI",
    "vendor": "customendpoint",
    "apiType": "responses",
    "models": [
      {
        "id": "gpt-5.5",
        "name": "RoutinAI GPT 5.5",
        "url": "https://api.routin.ai/v1/responses?token={ApiKey}",
        "toolCalling": true,
        "vision": true,
        "thinking": true,
        "maxInputTokens": 200000,
        "maxOutputTokens": 32000,
        "supportsReasoningEffort": [
          "none",
          "minimal",
          "low",
          "medium",
          "high",
          "xhigh"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "gpt-5.4",
        "name": "RoutinAI GPT 5.4",
        "url": "https://api.routin.ai/v1/responses?token={ApiKey}",
        "toolCalling": true,
        "vision": true,
        "thinking": true,
        "maxInputTokens": 1050000,
        "maxOutputTokens": 128000,
        "supportsReasoningEffort": [
          "none",
          "minimal",
          "low",
          "medium",
          "high",
          "xhigh"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "gpt-5.4-mini",
        "name": "RoutinAI GPT 5.4 Mini",
        "url": "https://api.routin.ai/v1/responses?token={ApiKey}",
        "toolCalling": true,
        "vision": true,
        "thinking": true,
        "maxInputTokens": 400000,
        "maxOutputTokens": 128000,
        "supportsReasoningEffort": [
          "none",
          "minimal",
          "low",
          "medium",
          "high",
          "xhigh"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "claude-opus-4-7",
        "name": "RoutinAI Claude Opus 4.7",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "vision": true,
        "thinking": true,
        "maxInputTokens": 200000,
        "apiType": "messages",
        "maxOutputTokens": 32000,
        "supportsReasoningEffort": [
          "minimal",
          "low",
          "medium",
          "high"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "claude-opus-4-8",
        "name": "RoutinAI Claude Opus 4.8",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "vision": true,
        "thinking": true,
        "maxInputTokens": 200000,
        "apiType": "messages",
        "maxOutputTokens": 32000,
        "supportsReasoningEffort": [
          "minimal",
          "low",
          "medium",
          "high"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "mimo-v2.5",
        "name": "RoutinAI MiMo V2.5",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "apiType": "messages",
        "vision": true,
        "thinking": true,
        "maxInputTokens": 1000000,
        "maxOutputTokens": 32000,
        "supportsReasoningEffort": [
          "low",
          "medium",
          "high"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "mimo-v2.5-pro",
        "name": "RoutinAI MiMo V2.5 Pro",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "apiType": "messages",
        "vision": true,
        "thinking": true,
        "maxInputTokens": 1000000,
        "maxOutputTokens": 32000,
        "supportsReasoningEffort": [
          "low",
          "medium",
          "high"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "deepseek-v4-pro",
        "name": "RoutinAI DeepSeek V4 Pro",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "apiType": "messages",
        "vision": false,
        "thinking": true,
        "maxInputTokens": 1000000,
        "maxOutputTokens": 384000,
        "supportsReasoningEffort": [
          "high",
          "max"
        ],
        "reasoningEffortFormat": "responses"
      },
      {
        "id": "deepseek-v4-flash",
        "name": "RoutinAI DeepSeek V4 Flash",
        "url": "https://api.routin.ai/v1/messages?token={ApiKey}",
        "toolCalling": true,
        "apiType": "messages",
        "vision": false,
        "thinking": true,
        "maxInputTokens": 1000000,
        "maxOutputTokens": 384000,
        "supportsReasoningEffort": [
          "high",
          "max"
        ],
        "reasoningEffortFormat": "responses"
      }
    ]
  }
]

After saving the configuration, return to Copilot Chat on the right. The model list should show models whose names start with RoutinAI, as shown below.

Verify the Configuration

Select a RoutinAI model from the model list and send a simple test message, such as asking it to explain the current file or generate sample code. If the model replies normally, the configuration is active and you can start using Copilot for development.

The following image shows an example of using Copilot with a RoutinAI model.

Troubleshooting

If RoutinAI does not appear in the model list, confirm that the JSON is valid, save the file, and reopen Copilot Chat.
If requests fail, check whether your API key is valid and whether your network can access https://api.routin.ai.
If a model cannot process images, choose a model whose vision value is true.

Use VS Code Copilot

On this page