Configure HPI-hosted AI models¶
This guide explains how to use HPI-hosted language models with GitHub Copilot Chat and OpenCode.
Note
This feature is experimental and subject to change. Available models and configuration options may change over time. Model availability is not guaranteed.
Create an API key¶
- Open 101010.hpi.de.
- Navigate to Settings → Account → API keys.
- Create a new API key.
- Copy the key. You will use it when configuring your AI client.
GitHub Copilot Chat¶
Add the custom model¶
Follow the GitHub Copilot documentation on custom endpoints to add a custom model. Select “Chat Completions” as the API type.
The following configuration is a starting point — adjust token limits and other settings as needed:
{
"id": "HPI-Intern-Qwen3.6-35B-A3B",
"name": "HPI Qwen 3.6 35B",
"url": "https://101010.hpi.de/api/v1/chat/completions",
"apiType": "chat-completions",
"toolCalling": true,
"vision": false,
"thinking": true,
"streaming": true,
"maxInputTokens": 229376,
"maxOutputTokens": 32768
},
{
"id": "HPI-Intern-Qwen38-27B",
"name": "HPI Qwen 3.8 27B",
"url": "https://101010.hpi.de/api/v1/chat/completions",
"apiType": "chat-completions",
"toolCalling": true,
"vision": true,
"thinking": true,
"streaming": true,
"maxInputTokens": 467232,
"maxOutputTokens": 32768,
"supportsReasoningEffort": ["low", "medium"],
"reasoningEffortFormat": "chat-completions"
},
Start using the model¶
Select the configured model from the model picker in GitHub Copilot Chat.
OpenCode¶
OpenCode can use the same HPI endpoint. Add the following configuration to your OpenCode configuration file:
{
"$schema": "https://opencode.ai/config.json",
"lsp": true,
"provider": {
"hpi": {
"npm": "@ai-sdk/openai-compatible",
"name": "HPI",
"options": {
"baseURL": "https://101010.hpi.de/api/v1"
},
"models": {
"HPI-Intern-Qwen3.6-35B-A3B": {
"name": "HPI Qwen 3.6"
},
"HPI-Intern-Qwen38-27B": {
"name": "HPI Qwen 3.8",
"reasoning": true,
"interleaved": {
"field": "reasoning_content"
},
"variants": {
"low": {
"reasoningEffort": "low"
},
"medium": {
"reasoningEffort": "medium"
}
}
}
}
}
},
"enabled_providers": ["hpi"],
"model": "hpi/HPI-Intern-Qwen38-27B"
}