Multi-Vendors

This document explains how to configure different AI vendors in the ADP Chat Client.

📋 Configuration Location

All vendor configurations must be added to the APP_CONFIGS field in your .env file.

The APP_CONFIGS field contains a JSON array of vendor configurations:

# In server/.env file
APP_CONFIGS='[
    {
        "Vendor": "Tencent",
        "ApplicationId": "your-app-id",
        "Comment": "Health Assistant",
        "AppKey": "your-app-key",
        "International": false
    },
    {
        "Vendor": "Ollama",
        "ApplicationId": "ollama-gemma3-1b",
        "DisplayName": "Gemma 3 Local Model",
        "BaseUrl": "http://localhost:11434/v1",
        "ModelName": "gemma3:1b",
        "Temperature": 0.7,
        "MaxTokens": 2000
    },
    {
        "Vendor": "OpenAI", 
        "ApplicationId": "openai-gpt4",
        "DisplayName": "ChatGPT 4",
        "ApiKey": "sk-proj-...",
        "BaseUrl": "https://api.openai.com/v1",
        "ModelName": "gpt-4"
    }
]'

Supported Vendors

Vendor	Description	Use Case
Tencent	Tencent Cloud ADP	Enterprise agent developing platform
Ollama	Local LLM models	Privacy-focused, offline AI (Llama, Gemma, Mistral)
OpenAI	OpenAI GPT models	State-of-the-art language models

1. Tencent ADP

Use Case: Enterprise AI with knowledge base and RAG support

Configuration

{
    "Vendor": "Tencent",
    "ApplicationId": "1979154564259682304",
    "Comment": "Health Assistant",
    "AppKey": "your-app-key-here",
    "International": false
}

Required Fields

Field	Description	How to Obtain
`ApplicationId`	Tencent ADP application ID	ADP Console
`AppKey`	Application authentication key	ADP Console → App Settings
`International`	Region setting	`false` (China) / `true` (International)

Additional Setup

Configure Tencent Cloud credentials in .env:

TC_SECRET_APPID=your-appid
TC_SECRET_ID=your-secret-id
TC_SECRET_KEY=your-secret-key

Get credentials from:

China mainland: https://console.cloud.tencent.com/cam/capi
International: https://console.tencentcloud.com/cam/capi

2. Ollama

Use Case: Local, privacy-focused AI models (Llama, Gemma, Mistral)

Quick Start

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Pull a model
ollama pull gemma3:1b

# Start service
ollama serve

Configuration

{
    "Vendor": "Ollama",
    "ApplicationId": "ollama-gemma3-1b",
    "DisplayName": "Gemma 3 Local Model",
    "BaseUrl": "http://localhost:11434/v1",
    "ModelName": "gemma3:1b",
    "Temperature": 0.7,
    "MaxTokens": 2000
}

Required Fields

Field	Description	Default
`BaseUrl`	Ollama API endpoint (must end with `/v1`)	`http://localhost:11434/v1`
`ModelName`	Model identifier	`llama2`

Optional Parameters

Field	Description	Default	Range
`Temperature`	Randomness/creativity	`0.7`	`0.0` - `2.0`
`MaxTokens`	Max response length	`2000`	`1` - `8192`
`TopP`	Nucleus sampling	`1.0`	`0.1` - `1.0`
`FrequencyPenalty`	Reduce repetition	`0.0`	`-2.0` - `2.0`
`PresencePenalty`	Encourage new topics	`0.0`	`-2.0` - `2.0`

Popular Models

ollama pull llama2          # Llama 2 7B
ollama pull llama3.1:8b     # Llama 3.1 8B
ollama pull mistral:7b      # Mistral 7B
ollama pull gemma3:1b       # Gemma 1B

3. OpenAI

Use Case: State-of-the-art language models (GPT-3.5, GPT-4, GPT-5)

Configuration

{
    "Vendor": "OpenAI",
    "ApplicationId": "openai-gpt4",
    "DisplayName": "ChatGPT 4",
    "ApiKey": "sk-proj-...",
    "BaseUrl": "https://api.openai.com/v1",
    "ModelName": "gpt-4"
}

Required Fields

Field	Description	Example
`ApiKey`	OpenAI API key	`sk-proj-...`
`BaseUrl`	API endpoint	`https://api.openai.com/v1`
`ModelName`	Model identifier	`gpt-4`, `gpt-3.5-turbo`

Supported Models

Model	Use Case
`gpt-3.5-turbo`	Fast, cost-effective
`gpt-4`	Complex reasoning, coding
`gpt-4-turbo`	Balanced performance
`gpt-4.1`	Latest GPT-4 variant
`gpt-5`	Advanced reasoning (no custom temperature)

Special Note: GPT-5/o1/o3

These models don't support custom Temperature:

{
    "Vendor": "OpenAI",
    "ApplicationId": "openai-gpt5",
    "DisplayName": "GPT-5",
    "ApiKey": "sk-proj-...",
    "ModelName": "gpt-5"
    // Do NOT set Temperature
}

Configuration Examples

Single Vendor

APP_CONFIGS='[
    {
        "Vendor": "Ollama",
        "ApplicationId": "ollama-llama2",
        "DisplayName": "Llama 2",
        "BaseUrl": "http://localhost:11434/v1",
        "ModelName": "llama2"
    }
]'

Multiple Vendors

APP_CONFIGS='[
    {
        "Vendor": "Tencent",
        "ApplicationId": "1234567890",
        "Comment": "Health Assistant",
        "AppKey": "...",
        "International": false
    },
    {
        "Vendor": "Ollama",
        "ApplicationId": "ollama-gemma3",
        "DisplayName": "Gemma 3 Local",
        "BaseUrl": "http://localhost:11434/v1",
        "ModelName": "gemma3:1b"
    },
    {
        "Vendor": "OpenAI",
        "ApplicationId": "openai-gpt4",
        "DisplayName": "ChatGPT 4",
        "ApiKey": "sk-proj-...",
        "ModelName": "gpt-4"
    }
]'

Common Issues

Vendor not found

Check Vendor field is capitalized: "Tencent", "Ollama", "OpenAI"

Ollama connection failed

ollama serve  # Start Ollama service
curl http://localhost:11434/v1/models  # Test API

OpenAI API error 401

Verify API key at https://platform.openai.com/api-keys
Check account has credits

OpenAI temperature error

Remove Temperature field for GPT-5/o1/o3 models

Resources

Tencent ADP: https://adp.cloud.tencent.com/
Ollama: https://ollama.ai/
OpenAI: https://platform.openai.com/

Multi-Vendors

On this page