> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tokenlab.sh/llms.txt
> Use this file to discover all available pages before exploring further.
# AI Chat Applications
> Configure popular AI chat applications to use TokenLab's OpenAI-compatible API
## Overview
Most general AI chat apps connect to TokenLab through the standard OpenAI-compatible chat path.
In practice that means:
* use `https://api.tokenlab.sh/v1` as the base URL unless the app explicitly asks for a full endpoint URL
* use your TokenLab API key as the OpenAI-style API key
* pick models that the app can use in a standard chat workflow
General chat apps are usually **chat-completions oriented**, not Codex Responses WebSocket clients. For Codex CLI/App, Cursor, or other coding tools that need Responses-specific behavior, use the dedicated integration guides instead.
**Type**: Chat App
**Primary Path**: OpenAI-compatible chat-completions
**Support Confidence**: Supported path
## Supported Applications
| Application | Platform | Notes |
| ------------- | ------------------------- | ------------------------------------------------------------- |
| Open WebUI | Web, Self-hosted | Popular self-hosted choice with OpenAI-compatible connections |
| LibreChat | Web, Self-hosted | Multi-provider chat workspace with custom OpenAI endpoints |
| LobeChat | Web, Desktop, Self-hosted | Strong custom-provider support |
| Cherry Studio | Windows, macOS, Linux | Good OpenAI-compatible provider flow |
| Chatbox | Windows, macOS, Linux | Simple OpenAI-compatible setup |
| DeepChat | Windows, macOS, Linux | Desktop client with OpenAI-compatible provider settings |
| Jan | Desktop, Self-hosted | Local-first app with remote OpenAI-compatible providers |
| TypingMind | Web, Desktop | Uses a custom endpoint style |
| NextChat | Web, Self-hosted | Lightweight ChatGPT-style client with base-URL overrides |
| ChatWise | Windows, macOS | Standard custom provider pattern |
## Default Connection Values
| Field | Value |
| -------- | ---------------------------- |
| Base URL | `https://api.tokenlab.sh/v1` |
| API Key | `sk-your-tokenlab-key` |
| API Type | OpenAI Compatible |
If an app asks for a **full endpoint URL** instead of a base URL, use:
* Chat Completions: `https://api.tokenlab.sh/v1/chat/completions`
## Example App Setups
**Self-hosted LobeChat**
```bash theme={null}
OPENAI_API_KEY=sk-your-tokenlab-key
OPENAI_PROXY_URL=https://api.tokenlab.sh/v1
```
**LobeChat Cloud**
1. Open **Settings** → **AI Service Provider**
2. Select **OpenAI**
3. Enter your TokenLab API key
4. Set the custom endpoint to `https://api.tokenlab.sh/v1`
1. Open **Settings** → **Model Provider**
2. Add an **OpenAI-compatible** provider
3. Enter:
* **Name**: TokenLab
* **API Base URL**: `https://api.tokenlab.sh/v1`
* **API Key**: `sk-your-tokenlab-key`
4. Add the models you want to use
1. Open **Settings** → **AI Model**
2. Choose **OpenAI API Compatible**
3. Enter:
* **API Host**: `https://api.tokenlab.sh/v1`
* **API Key**: `sk-your-tokenlab-key`
* **Model**: `gpt-5.4-mini` or `claude-sonnet-5`
1. Open **Settings**
2. Choose **Custom Endpoint**
3. Enter:
* **Endpoint URL**: `https://api.tokenlab.sh/v1/chat/completions`
* **API Key**: `sk-your-tokenlab-key`
4. Configure the models you want to expose
1. Open **Admin Settings** → **Connections**
2. Add an **OpenAI-compatible** connection
3. Enter:
* **Base URL**: `https://api.tokenlab.sh/v1`
* **API Key**: `sk-your-tokenlab-key`
4. Save and select the models you want to expose
1. Open **Settings** -> **Model Provider**
2. Add an OpenAI-compatible provider
3. Enter:
* **Base URL**: `https://api.tokenlab.sh/v1`
* **API Key**: `sk-your-tokenlab-key`
* **Model**: `gpt-5.4-mini`
4. Save and run a test chat
1. Open **Settings** -> **Providers**
2. Add a remote OpenAI-compatible provider
3. Enter:
* **Provider name**: TokenLab
* **Base URL**: `https://api.tokenlab.sh/v1`
* **API Key**: `sk-your-tokenlab-key`
4. Select `gpt-5.4-mini` or another TokenLab model
Add an OpenAI-compatible endpoint in your LibreChat provider config and point it to:
```yaml theme={null}
customEndpoints:
- name: TokenLab
apiKey: '${OPENAI_API_KEY}'
baseURL: 'https://api.tokenlab.sh/v1'
models:
default: ['gpt-5.4-mini', 'gpt-5.4', 'claude-sonnet-5']
fetch: false
```
## Safer Model Defaults
| Category | Suggested Models |
| --------------- | ------------------------------------------------------ |
| General default | `gpt-5.4-mini`, `gpt-5.4`, `gpt-5.5` |
| Anthropic | `claude-sonnet-5`, `claude-opus-4-8`, `claude-fable-5` |
| Google | `gemini-3.5-flash`, `gemini-3.1-flash-lite` |
| DeepSeek | `deepseek-v4-flash`, `deepseek-v4-pro` |
## Tips
Most apps are easier to configure and less fragile when you provide `https://api.tokenlab.sh/v1` as a base URL instead of a full endpoint URL.
Start with `gpt-5.4-mini` or `deepseek-v4-flash` for everyday chat, then move heavier reasoning tasks to `gpt-5.4` or `claude-opus-4-8`.
General chat apps and coding agents often need different endpoint behavior. Keep Codex/Cursor-style tools on their dedicated integration pages.
## Troubleshooting
* Verify the base URL is exactly `https://api.tokenlab.sh/v1`
* Remove trailing slashes if the app auto-appends paths
* If the app wants a full endpoint URL, use `https://api.tokenlab.sh/v1/chat/completions`
* Double-check your API key
* Confirm the key is active in TokenLab dashboard
* Make sure there are no extra spaces or line breaks
* Verify the model name exactly
* Try a widely supported default like `gpt-5.4-mini`, `deepseek-v4-flash`, or `claude-sonnet-5`
* Check current availability in TokenLab docs or dashboard