Instance Configuration Examples¶

This document provides configuration examples for common AI services to help users get started quickly.

Online API Services¶

OpenAI GPT¶

Field	Value
Service Type	`chat`
Instance Name	`gpt-4o`
Base URL	`https://api.openai.com`
Path	`/v1/chat/completions`
Adapter	`normal`
Headers	`Authorization: Bearer sk-xxx`

Base URL:     https://api.openai.com
Path:         /v1/chat/completions
Header Name:  Authorization
Header Value: Bearer sk-proj-xxxxx

DeepSeek¶

DeepSeek is compatible with OpenAI API format, use normal adapter.

Field	Value
Service Type	`chat`
Instance Name	`deepseek-chat`
Base URL	`https://api.deepseek.com`
Path	`/v1/chat/completions`
Adapter	`normal`
Headers	`Authorization: Bearer sk-xxx`

Base URL:     https://api.deepseek.com
Path:         /v1/chat/completions
Header Name:  Authorization
Header Value: Bearer sk-xxxxxxxx

Anthropic Claude¶

Claude uses a dedicated adapter with x-api-key header authentication.

Field	Value
Service Type	`chat`
Instance Name	`claude-sonnet`
Base URL	`https://api.anthropic.com`
Path	`/v1/messages`
Adapter	`claude`
Headers	`x-api-key: sk-ant-xxx`

Base URL:     https://api.anthropic.com
Path:         /v1/messages
Header Name:  x-api-key
Header Value: sk-ant-api03-xxxxx

Note: The anthropic-version header is automatically injected by the adapter.

Google Gemini¶

Gemini is compatible with OpenAI API format, use normal adapter.

Field	Value
Service Type	`chat`
Instance Name	`gemini-pro`
Base URL	`https://generativelanguage.googleapis.com/v1beta/openai`
Path	`/chat/completions`
Adapter	`normal`
Headers	`Authorization: Bearer xxx`

Base URL:     https://generativelanguage.googleapis.com/v1beta/openai
Path:         /chat/completions
Header Name:  Authorization
Header Value: Bearer AIzaSyxxxxx

Alibaba Cloud Bailian (Qwen)¶

Field	Value
Service Type	`chat`
Instance Name	`qwen-plus`
Base URL	`https://dashscope.aliyuncs.com/compatible-mode`
Path	`/v1/chat/completions`
Adapter	`normal`
Headers	`Authorization: Bearer sk-xxx`

Base URL:     https://dashscope.aliyuncs.com/compatible-mode
Path:         /v1/chat/completions
Header Name:  Authorization
Header Value: Bearer sk-xxxxxxxx

Tencent Cloud Hunyuan¶

Field	Value
Service Type	`chat`
Instance Name	`hunyuan-chat`
Base URL	`https://api.hunyuan.cloud.tencent.com`
Path	`/v1/chat/completions`
Adapter	`normal`
Headers	`Authorization: Bearer sk-xxx`

Base URL:     https://api.hunyuan.cloud.tencent.com
Path:         /v1/chat/completions
Header Name:  Authorization
Header Value: Bearer sk-xxxxxxxx

Local Inference Engines¶

Ollama¶

Field	Value
Service Type	`chat`
Instance Name	`ollama-llama3`
Base URL	`http://localhost:11434`
Path	`/api/chat`
Adapter	`ollama`
Headers	None required

Base URL:     http://localhost:11434
Path:         /api/chat
Adapter:      ollama
Headers:      None required

vLLM¶

Field	Value
Service Type	`chat`
Instance Name	`vllm-qwen`
Base URL	`http://localhost:8000`
Path	`/v1/chat/completions`
Adapter	`vllm`
Headers	None required (unless API key configured)

Base URL:     http://localhost:8000
Path:         /v1/chat/completions
Adapter:      vllm
Headers:      None required

GPUStack¶

Field	Value
Service Type	`chat`
Instance Name	`gpustack-llama`
Base URL	`http://localhost:8080`
Path	`/v1/chat/completions`
Adapter	`gpustack`
Headers	`Authorization: Bearer sk-xxx` (if configured)

Base URL:     http://localhost:8080
Path:         /v1/chat/completions
Adapter:      gpustack
Header Name:  Authorization (optional)
Header Value: Bearer your-api-key (optional)

Xinference¶

Field	Value
Service Type	`chat`
Instance Name	`xinference-chat`
Base URL	`http://localhost:9997`
Path	`/v1/chat/completions`
Adapter	`xinference`
Headers	None required

Base URL:     http://localhost:9997
Path:         /v1/chat/completions
Adapter:      xinference
Headers:      None required

LocalAI¶

Field	Value
Service Type	`chat`
Instance Name	`localai-chat`
Base URL	`http://localhost:8080`
Path	`/v1/chat/completions`
Adapter	`localai`
Headers	None required

Base URL:     http://localhost:8080
Path:         /v1/chat/completions
Adapter:      localai
Headers:      None required

Adapter Reference¶

Adapter	Supported Services	Authentication
`normal`	OpenAI, DeepSeek, Gemini, Alibaba, Tencent	`Authorization: Bearer`
`claude`	Anthropic Claude	`x-api-key`
`ollama`	Ollama	None
`vllm`	vLLM	None (optional)
`gpustack`	GPUStack	`Authorization: Bearer` (optional)
`xinference`	Xinference	None
`localai`	LocalAI	None

Embedding Service Examples¶

OpenAI Embedding¶

Field	Value
Service Type	`embedding`
Instance Name	`text-embedding-3-small`
Base URL	`https://api.openai.com`
Path	`/v1/embeddings`
Adapter	`normal`
Headers	`Authorization: Bearer sk-xxx`

Ollama Embedding¶

Field	Value
Service Type	`embedding`
Instance Name	`nomic-embed-text`
Base URL	`http://localhost:11434`
Path	`/api/embeddings`
Adapter	`ollama`
Headers	None required

FAQ¶

Q: How to configure multiple OpenAI instances?¶

A: Add multiple instances with different names and weights. For example:

Instance 1: gpt-4o        Weight 5
Instance 2: gpt-4o-mini   Weight 3
Instance 3: gpt-3.5-turbo Weight 2

Q: Will the API Key in headers be exposed?¶

A: API Keys are displayed as password fields on the page and transmitted over HTTPS. JAiRouter masks stored keys internally.

Q: Do I need to manually configure the `anthropic-version` header for Claude?¶

A: No. When you select the claude adapter, anthropic-version: 2023-06-01 is automatically injected.

Q: Local inference engines don't require authentication, do I need to configure headers?¶

A: No. Ollama, vLLM, Xinference and other local services don't require authentication by default, leave headers empty.