API Configuration

Through “Model Settings” → “Add API”, you can connect any compatible model service to NextAI.
Supports both custom APIs and common providers such as Qwen, Gemini, Anthropic, and Ollama.

1. Creating a New API Provider

1. Open NextAI → Tap the top floating island → Open the menu → Go to Settings → API Settings.
2. Tap the “＋” icon at the top right to create a new API.

You’ll see a form with the following fields:

Field	Description
Enable	When turned on, this provider will be available in chat.
Type	Select model type: Custom API / Qwen / Gemini / Anthropic / Ollama.
Display Name	Custom label for the provider (e.g., OpenAI, DeepSeek, or My Gateway).
API Address (Base URL)	The endpoint for model requests, e.g., https://api.openai.com/v1.
API Key	Copy your API key from the corresponding provider’s console.

💡 Tip:

• “Custom API” is compatible with most OpenAI-style APIs.
• Selecting a specific provider (like Qwen or Gemini) may auto-fill the default address.

After completing the form, tap 💾 Save at the top right.

2. Common Provider Examples

Provider	Base URL Example
OpenAI	https://api.openai.com/v1
DeepSeek	https://api.deepseek.com
Qwen (Tongyi)	https://dashscope.aliyuncs.com/compatible-mode/v1
Moonshot	https://api.moonshot.cn/v1
Zhipu (GLM)	https://open.bigmodel.cn/api/paas/v4
Ollama	Local deployment, e.g. http://127.0.0.1:11434/api

3. Save and Use

1. Tap Save, then return to the chat page.
2. You can now select the newly added provider when starting a new conversation.
   If prompted with “Please add a model provider first,” check:
   • The provider is enabled;
   • Base URL and Key are correct;
   • The network can reach the domain (some networks block external APIs).

4. Security Recommendations

• API Keys are stored locally on your device and never uploaded to any server.
• Use separate keys for each device and rotate them regularly.
• Double-check for spaces or newlines when pasting keys.
• Avoid putting proxy URLs directly into Base URL fields.

5. Test Connectivity (Optional)

You can test your API connection using curl:

curl https://api.openai.com/v1/models \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json"

If you receive a JSON model list, your connection works.
If you get 401 / 403 / 404 / 5xx errors, refer to the troubleshooting guide.

6. Add and Enable Models

After adding an API, you must configure one or more models under it.

1. Go to Settings → API Settings.
2. Under the target provider (e.g., Qwen), tap “＋ Add Model”.
3. Fill in the form as follows:

Field	Description
Enable	Enable to make the model available for chatting.
Multimodal	Turn on if the model supports image or voice input.
Display Name	Shown in chat (e.g., plus, turbo, glm-4).
Max Token Count	Controls context length (default: 2048).
Model ID	Required — the model’s unique ID (e.g., gpt-4-turbo, qwen-turbo, gemini-pro).

Tap 💾 Save when finished.

Common Model IDs

Provider	Model ID	Description
OpenAI	gpt-4-turbo, gpt-3.5-turbo	Supports multimodal inputs, strong general performance
DeepSeek	deepseek-chat, deepseek-coder	Optimized for Chinese & coding
Qwen	qwen-turbo, qwen-plus	Tongyi Qwen family models
Gemini	gemini-pro	Google multimodal model
Anthropic	claude-3-opus, claude-3-haiku	Strong English reasoning
Ollama	llama3, mistral, phi3	Local models, offline support

Tips

• Each API can have multiple models for quick switching.
• Models are one-to-many with APIs; deleting an API removes its models.
• You can configure multiple APIs for the same provider (e.g., with different gateways).
• For multimodal models (like GPT-4 Vision or Gemini Pro), enable the Multimodal toggle.

Proxy & Network Tips

• If you use a local proxy (e.g. 127.0.0.1:7890), ensure your system settings allow NextAI to access it.
• Corporate or campus networks may restrict outbound API requests — try a home or mobile network.
• Don’t put proxy addresses in Base URL fields — always use the provider’s official API domain.