LLM Provider Configuration

NekoBot supports multiple Large Language Model (LLM) providers, including OpenAI, Anthropic, Google Gemini, and more.

Configuration Overview

Configure LLM providers in the provider_configs object in data/config.json:

json

{
  "provider_configs": {
    "openai": {
      "api_key": "sk-your-openai-api-key",
      "default_model": "gpt-4o"
    },
    "anthropic": {
      "api_key": "sk-ant-your-anthropic-api-key",
      "default_model": "claude-3-5-sonnet-20241022"
    },
    "gemini": {
      "api_key": "your-google-api-key",
      "default_model": "gemini-1.5-pro"
    }
  }
}

Supported Providers

1. OpenAI

Supports OpenAI official API and OpenAI-compatible APIs.

Basic Configuration

json

{
  "openai": {
    "api_key": "sk-your-api-key",
    "base_url": "https://api.openai.com/v1",
    "default_model": "gpt-4o",
    "timeout": 30,
    "max_retries": 3
  }
}

Configuration Parameters

Parameter	Type	Required	Default	Description
`api_key`	string	Yes	-	OpenAI API key
`base_url`	string	No	`"https://api.openai.com/v1"`	API base URL
`default_model`	string	No	`"gpt-4o"`	Default model
`timeout`	integer	No	`30`	Request timeout (seconds)
`max_retries`	integer	No	`3`	Maximum retry attempts
`organization`	string	No	`null`	Organization ID
`project`	string	No	`null`	Project ID

Supported Models

gpt-5.4 - GPT-5.4 (latest version)
gpt-5.4-mini - GPT-5.4 Mini (lightweight)
gpt-5.4-pro - GPT-5.4 Pro (professional)
gpt-5.4-nano - GPT-5.4 Nano (ultra-lightweight)
gpt-5 - GPT-5
gpt-5-mini - GPT-5 Mini
gpt-5-pro - GPT-5 Pro
gpt-5.1 - GPT-5.1
gpt-5.2 - GPT-5.2
gpt-5.3 - GPT-5.3
gpt-4.1 - GPT-4.1 (reasoning model)
gpt-4.1-mini - GPT-4.1 Mini (lightweight reasoning)
gpt-4o - GPT-4o (multimodal model)
gpt-4o-mini - GPT-4o Mini (lightweight multimodal)
gpt-4-turbo - GPT-4 Turbo
gpt-4 - GPT-4
gpt-3.5-turbo - GPT-3.5 Turbo
o1-preview - OpenAI o1 preview (reasoning optimized)
o1-mini - OpenAI o1 mini (lightweight reasoning)

Note: Model list based on models.dev, actual availability depends on OpenAI API support. NekoBot automatically fetches model capabilities from models.dev.

2. Anthropic (Claude)

Supports Anthropic Claude series models.

Basic Configuration

json

{
  "anthropic": {
    "api_key": "sk-ant-your-api-key",
    "default_model": "claude-sonnet-4-6",
    "base_url": "https://api.anthropic.com",
    "timeout": 30
  }
}

Configuration Parameters

Parameter	Type	Required	Default	Description
`api_key`	string	Yes	-	Anthropic API key
`default_model`	string	No	`"claude-sonnet-4-6"`	Default model
`base_url`	string	No	`"https://api.anthropic.com"`	API base URL
`timeout`	integer	No	`30`	Request timeout (seconds)
`max_tokens`	integer	No	`4096`	Maximum generation tokens
`temperature`	number	No	`0.7`	Temperature parameter (0-1)

Supported Models

claude-3-7-sonnet-latest - Claude 3.7 Sonnet (latest version)
claude-sonnet-4-20250514 - Claude Sonnet 4
claude-opus-4-7 - Claude Opus 4.7 (strongest reasoning)
claude-sonnet-4-6 - Claude Sonnet 4.6 (best value)
claude-haiku-4-5 - Claude Haiku 4.5 (fastest)
claude-opus-4-6 - Claude Opus 4.6 (previous strongest)
claude-sonnet-4-5 - Claude Sonnet 4.5
claude-opus-4-5 - Claude Opus 4.5
claude-3-5-sonnet-20241022 - Claude 3.5 Sonnet (legacy)
claude-3-5-haiku-20241022 - Claude 3.5 Haiku (legacy)

3. Google Gemini

Supports Google Gemini series models.

Basic Configuration

json

{
  "gemini": {
    "api_key": "your-google-api-key",
    "default_model": "gemini-2.5-pro",
    "safety_settings": {
      "HARM_CATEGORY_HARASSMENT": "BLOCK_NONE",
      "HARM_CATEGORY_HATE_SPEECH": "BLOCK_NONE",
      "HARM_CATEGORY_SEXUALLY_EXPLICIT": "BLOCK_NONE",
      "HARM_CATEGORY_DANGEROUS_CONTENT": "BLOCK_NONE"
    }
  }
}

Configuration Parameters

Parameter	Type	Required	Default	Description
`api_key`	string	Yes	-	Google API key
`default_model`	string	No	`"gemini-2.5-pro"`	Default model
`safety_settings`	object	No	`{}`	Safety settings
`generation_config`	object	No	`{}`	Generation configuration
`timeout`	integer	No	`30`	Request timeout (seconds)

Supported Models

gemini-2.5-pro - Gemini 2.5 Pro (latest version)
gemini-2.5-flash - Gemini 2.5 Flash (fast version)
gemini-1.5-pro - Gemini 1.5 Pro
gemini-1.5-flash - Gemini 1.5 Flash
gemini-pro - Gemini Pro
gemini-ultra - Gemini Ultra

4. OpenAI Compatible API

Supports third-party services compatible with OpenAI API, such as DeepSeek, Ollama, etc.

Basic Configuration

json

{
  "openai_compatible": {
    "api_key": "your-api-key",
    "base_url": "https://api.deepseek.com/v1",
    "default_model": "deepseek-chat",
    "provider_name": "deepseek"
  }
}

Configuration Parameters

Parameter	Type	Required	Default	Description
`api_key`	string	Yes	-	API key
`base_url`	string	Yes	-	API base URL
`default_model`	string	No	`null`	Default model
`provider_name`	string	No	`"openai_compatible"`	Provider name
`timeout`	integer	No	`30`	Request timeout (seconds)

Supported Compatible Services

DeepSeek: https://api.deepseek.com/v1
Ollama: http://localhost:11434/v1
LocalAI: http://localhost:8080/v1
Together AI: https://api.together.xyz/v1
Azure OpenAI: https://{resource}.openai.azure.com/openai/deployments/{deployment}

5. Text-to-Speech (TTS)

Edge TTS

Uses Microsoft Edge's speech synthesis service.

json

{
  "edge_tts": {
    "voice": "zh-CN-XiaoxiaoNeural",
    "rate": "+0%",
    "volume": "+0%",
    "pitch": "+0Hz"
  }
}

OpenAI TTS

Uses OpenAI's text-to-speech service.

json

{
  "openai_tts": {
    "api_key": "sk-your-openai-api-key",
    "model": "tts-1",
    "voice": "alloy",
    "speed": 1.0
  }
}

6. Speech-to-Text (STT)

OpenAI Whisper

Uses OpenAI's speech recognition service.

json

{
  "openai_stt": {
    "api_key": "sk-your-openai-api-key",
    "model": "whisper-1",
    "language": "zh",
    "temperature": 0.0
  }
}

Multi-Provider Configuration

You can configure multiple providers simultaneously, and NekoBot will automatically select as needed:

json

{
  "provider_configs": {
    "openai": {
      "api_key": "sk-openai-key",
      "default_model": "gpt-4o"
    },
    "anthropic": {
      "api_key": "sk-ant-key",
      "default_model": "claude-3-5-sonnet"
    },
    "gemini": {
      "api_key": "google-key",
      "default_model": "gemini-1.5-pro"
    },
    "deepseek": {
      "api_key": "deepseek-key",
      "base_url": "https://api.deepseek.com/v1",
      "default_model": "deepseek-chat",
      "provider_name": "deepseek"
    }
  }
}

Provider Selection Strategy

1. Default Provider

Specify default provider in plugins or configuration:

python

# Specify provider in plugin
await self.context.request_provider("openai", prompt="Hello")

2. Fallback Strategy

If the primary provider fails, automatically try other available providers:

Try the first configured provider
If fails, try other providers in configuration order
Record failure reasons for debugging

3. Load Balancing

For multiple providers of the same type, configure load balancing:

json

{
  "provider_configs": {
    "openai_1": {
      "api_key": "key1",
      "weight": 0.5
    },
    "openai_2": {
      "api_key": "key2", 
      "weight": 0.5
    }
  }
}

Advanced Configuration

Model-Specific Configuration

Set specific parameters for different models:

json

{
  "openai": {
    "api_key": "sk-key",
    "models": {
      "gpt-4o": {
        "max_tokens": 4096,
        "temperature": 0.7
      },
      "gpt-3.5-turbo": {
        "max_tokens": 2048,
        "temperature": 0.8
      }
    }
  }
}

Proxy Configuration

Access API through proxy:

json

{
  "openai": {
    "api_key": "sk-key",
    "http_client": {
      "proxy": "http://proxy.example.com:8080",
      "verify_ssl": true
    }
  }
}

Request Limits

Configure request rate limiting:

json

{
  "openai": {
    "api_key": "sk-key",
    "rate_limit": {
      "requests_per_minute": 60,
      "tokens_per_minute": 90000
    }
  }
}

Environment Variable Configuration

Use environment variables instead of sensitive information in configuration files:

bash

export OPENAI_API_KEY="sk-your-key"
export ANTHROPIC_API_KEY="sk-ant-your-key"
export GEMINI_API_KEY="your-google-key"

Then reference environment variables in configuration file:

json

{
  "openai": {
    "api_key": "${OPENAI_API_KEY}"
  },
  "anthropic": {
    "api_key": "${ANTHROPIC_API_KEY}"
  }
}

Troubleshooting

API Key Errors

ERROR: Invalid API key provided

Solutions:

Check if API key is correct
Check for spaces or special characters
Check if key has expired

Network Connection Issues

ERROR: Connection timeout

Solutions:

Check network connection
Check firewall settings
Try increasing timeout parameter
Check proxy configuration

Model Unavailable

ERROR: Model not found

Solutions:

Check if model name is correct
Check if API supports the model
Check provider documentation for available model list

Quota Limits

ERROR: Rate limit exceeded

Solutions:

Reduce request frequency
Configure rate_limit parameter
Upgrade API plan
Use multiple providers to distribute requests

Testing Configuration

1. Validate Configuration

Automatically validate configuration when starting NekoBot:

bash

python -m nekobot

2. Test Provider Connection

Use Web UI to test provider connections:

Access http://localhost:6285
Go to provider management page
Click "Test Connection" button

3. Check Logs

Check NekoBot logs for detailed error information:

bash

# View real-time logs
tail -f logs/nekobot.log

Custom Providers

You can register custom providers through plugins, see Plugin Development documentation.

Next Steps

Framework Configuration - Configure permissions, moderation, and conversation management
Platform Configuration - Configure chat platforms
Plugin Development - Develop custom features
Quick Start - Review basic configuration

LLM Provider Configuration ​

Configuration Overview ​

Supported Providers ​

1. OpenAI ​

Basic Configuration ​

Configuration Parameters ​

Supported Models ​

2. Anthropic (Claude) ​

Basic Configuration ​

Configuration Parameters ​

Supported Models ​

3. Google Gemini ​

Basic Configuration ​

Configuration Parameters ​

Supported Models ​

4. OpenAI Compatible API ​

Basic Configuration ​

Configuration Parameters ​

Supported Compatible Services ​

5. Text-to-Speech (TTS) ​

Edge TTS ​

OpenAI TTS ​

6. Speech-to-Text (STT) ​

OpenAI Whisper ​

Multi-Provider Configuration ​

Provider Selection Strategy ​

1. Default Provider ​

2. Fallback Strategy ​

3. Load Balancing ​

Advanced Configuration ​

Model-Specific Configuration ​

Proxy Configuration ​

Request Limits ​

Environment Variable Configuration ​

Troubleshooting ​

API Key Errors ​

Network Connection Issues ​

Model Unavailable ​

Quota Limits ​

Testing Configuration ​

1. Validate Configuration ​

2. Test Provider Connection ​

3. Check Logs ​

Custom Providers ​

Next Steps ​

LLM Provider Configuration

Configuration Overview

Supported Providers

1. OpenAI

Basic Configuration

Configuration Parameters

Supported Models

2. Anthropic (Claude)

Basic Configuration

Configuration Parameters

Supported Models

3. Google Gemini

Basic Configuration

Configuration Parameters

Supported Models

4. OpenAI Compatible API

Basic Configuration

Configuration Parameters

Supported Compatible Services

5. Text-to-Speech (TTS)

Edge TTS

OpenAI TTS

6. Speech-to-Text (STT)

OpenAI Whisper

Multi-Provider Configuration

Provider Selection Strategy

1. Default Provider

2. Fallback Strategy

3. Load Balancing

Advanced Configuration

Model-Specific Configuration

Proxy Configuration

Request Limits

Environment Variable Configuration

Troubleshooting

API Key Errors

Network Connection Issues

Model Unavailable

Quota Limits

Testing Configuration

1. Validate Configuration

2. Test Provider Connection

3. Check Logs

Custom Providers

Next Steps