v0.5.2 release - Contributors, Sponsors and Enquiries are most welcome 😌

Provider Reference

Complete guide to all 12+ LLM providers supported by AgentSea ADK. Mix and match providers for your needs.

🔌 12+ Providers Supported

Cloud Providers (6): Anthropic, OpenAI, Google, Azure OpenAI, Mistral AI, Cohere

Local Providers (6): Ollama, LM Studio, LocalAI, Text Generation WebUI, vLLM, Jan

Voice Providers (7): OpenAI Whisper, LemonFox STT, Local Whisper, OpenAI TTS, LemonFox TTS, ElevenLabs, Piper TTS

Provider Categories

☁️

Cloud Providers

Hosted APIs with best quality, easy to use, pay-per-token pricing.

🏠

Local Providers

Self-hosted models, complete privacy, no API costs, offline capability.

🎙️

Voice Providers

Speech-to-Text and Text-to-Speech for voice-enabled agents.

Cloud Providers

Anthropic (Claude)

Leading AI safety company known for Claude models with strong reasoning and long context windows.

typescript
import { AnthropicProvider } from '@lov3kaizen/agentsea-core';

const provider = new AnthropicProvider(process.env.ANTHROPIC_API_KEY);

// Available models:
// - claude-sonnet-4-20250514 (Latest, most capable)
// - claude-3-5-sonnet-20241022
// - claude-3-opus-20240229 (Highest intelligence)
// - claude-3-haiku-20240307 (Fastest, cheapest)

const agent = new Agent(
  {
    model: 'claude-sonnet-4-20250514',
    provider: 'anthropic',
    temperature: 0.7,
    maxTokens: 4096,
  },
  provider,
  toolRegistry,
);

Strengths:

  • ✅ Excellent reasoning and analysis
  • ✅ 200K context window
  • ✅ Strong safety features
  • ✅ Tool use (function calling)
  • ✅ Vision capabilities

Best For:

  • Complex reasoning tasks
  • Code analysis
  • Long document processing
  • Safety-critical applications

OpenAI (GPT)

Pioneer in large language models, known for GPT-4 and strong ecosystem.

typescript
import { OpenAIProvider } from '@lov3kaizen/agentsea-core';

const provider = new OpenAIProvider(process.env.OPENAI_API_KEY);

// Available models:
// - gpt-4-turbo-preview (Most capable)
// - gpt-4 (Reliable, proven)
// - gpt-3.5-turbo (Fast, cost-effective)
// - gpt-4-vision-preview (Vision support)

const agent = new Agent(
  {
    model: 'gpt-4-turbo-preview',
    provider: 'openai',
    temperature: 0.8,
    maxTokens: 2048,
  },
  provider,
  toolRegistry,
);

Strengths:

  • ✅ Broad knowledge base
  • ✅ Strong creative writing
  • ✅ Function calling
  • ✅ Large ecosystem
  • ✅ Reliable performance

Best For:

  • General-purpose tasks
  • Creative content
  • Customer service
  • Content generation

Google (Gemini)

Google's multimodal AI models with strong reasoning and native tool integration.

typescript
import { GoogleProvider } from '@lov3kaizen/agentsea-core';

const provider = new GoogleProvider(process.env.GOOGLE_AI_API_KEY);

// Available models:
// - gemini-2.0-flash-exp (Latest, fastest)
// - gemini-1.5-pro (Most capable)
// - gemini-1.5-flash (Fast, affordable)

const agent = new Agent(
  {
    model: 'gemini-2.0-flash-exp',
    provider: 'google',
    temperature: 0.7,
  },
  provider,
  toolRegistry,
);

Strengths:

  • ✅ Multimodal (text, image, video, audio)
  • ✅ 1M+ context window
  • ✅ Fast inference
  • ✅ Native function calling
  • ✅ Code execution

Best For:

  • Multimodal applications
  • Long context tasks
  • Video/audio analysis
  • Scientific research

Other Cloud Providers

Azure OpenAI

Enterprise-grade OpenAI models with Azure infrastructure and compliance.

AzureOpenAIProvider

Mistral AI

European AI company with strong open models and competitive pricing.

MistralAIProvider

Cohere

Enterprise-focused AI with strong retrieval and generation capabilities.

CohereProvider

Local Providers

🔒 Why Local?

✅ Privacy: Data never leaves your infrastructure

✅ Cost: Zero API costs, unlimited usage

✅ Control: Full control over models and versions

✅ Offline: Works without internet connection

Ollama (Recommended)

The easiest way to run models locally. Simple CLI, automatic GPU support, growing model library.

typescript
import { OllamaProvider } from '@lov3kaizen/agentsea-core';

const provider = new OllamaProvider({
  baseUrl: 'http://localhost:11434',
});

// Pull models
await provider.pullModel('llama3.2');
await provider.pullModel('mistral');

// List available models
const models = await provider.listModels();

const agent = new Agent(
  {
    model: 'llama3.2',
    provider: 'ollama',
  },
  provider,
  toolRegistry,
);

Popular Models: llama3.2 (8B), mistral (7B), qwen2.5 (7B), gemma2 (9B), codellama (7B), phi3 (3.8B)

LM Studio

Desktop app with beautiful UI. Download models with one click, built-in OpenAI-compatible server.

typescript
import { LMStudioProvider } from '@lov3kaizen/agentsea-core';

const provider = new LMStudioProvider({
  baseUrl: 'http://localhost:1234',
});

const agent = new Agent(
  {
    model: 'local-model', // Whatever you loaded in LM Studio
    provider: 'lm-studio',
  },
  provider,
  toolRegistry,
);

LocalAI

Self-hosted OpenAI alternative. Supports LLMs, Stable Diffusion, voice, embeddings, more.

typescript
import { LocalAIProvider } from '@lov3kaizen/agentsea-core';

const provider = new LocalAIProvider({
  baseUrl: 'http://localhost:8080',
});

const agent = new Agent(
  {
    model: 'llama-3.2-3b',
    provider: 'localai',
  },
  provider,
  toolRegistry,
);

Text Generation WebUI

Feature-rich web interface for running models. Extensions, characters, multiple backends.

typescript
import { TextGenerationWebUIProvider } from '@lov3kaizen/agentsea-core';

const provider = new TextGenerationWebUIProvider({
  baseUrl: 'http://localhost:5000',
});

vLLM

High-throughput inference engine for production. Uses PagedAttention for efficiency.

typescript
import { VLLMProvider } from '@lov3kaizen/agentsea-core';

const provider = new VLLMProvider({
  baseUrl: 'http://localhost:8000',
});

// Best for production with high request volume

Jan

Open source ChatGPT alternative. Desktop app with local execution.

typescript
import { OpenAICompatibleProvider } from '@lov3kaizen/agentsea-core';

// Jan uses OpenAI-compatible API
const provider = new OpenAICompatibleProvider({
  baseUrl: 'http://localhost:1337',
});

Provider Comparison

Cloud Providers

ProviderQualitySpeedCostContextBest For
Anthropic⭐⭐⭐⭐⭐⭐⭐⭐⭐$$$200KReasoning, safety
OpenAI⭐⭐⭐⭐⭐⭐⭐⭐⭐$$$128KGeneral purpose
Google⭐⭐⭐⭐⭐⭐⭐⭐⭐⭐$$1M+Multimodal, long context
Mistral AI⭐⭐⭐⭐⭐⭐⭐⭐$$32KEuropean compliance
Cohere⭐⭐⭐⭐⭐⭐⭐⭐$$128KEnterprise RAG

Local Providers

ProviderEase of UsePerformanceFeaturesBest For
Ollama⭐⭐⭐⭐⭐⭐⭐⭐⭐Model mgmt, CLIGetting started
LM Studio⭐⭐⭐⭐⭐⭐⭐⭐⭐GUI, easy setupNon-technical users
LocalAI⭐⭐⭐⭐⭐⭐⭐Multi-modal, DockerSelf-hosted services
vLLM⭐⭐⭐⭐⭐⭐⭐PagedAttentionProduction scale
Text Gen WebUI⭐⭐⭐⭐⭐⭐⭐Web UI, extensionsExperimentation
Jan⭐⭐⭐⭐⭐⭐⭐⭐Desktop appChatGPT alternative

Choosing a Provider

🎯 For Getting Started

Cloud: Start with Anthropic Claude or OpenAI GPT - easy setup, excellent quality
Local: Ollama with llama3.2 - simplest local setup

💰 For Cost Savings

Development: Ollama (free, unlimited)
Production: Google Gemini Flash (lowest cost per token) or vLLM (self-hosted)

🔒 For Privacy

Complete Privacy: Ollama + Local Whisper + Piper TTS
Production Scale: vLLM with self-hosted models

⚡ For Performance

Speed: Google Gemini Flash or Claude Haiku
Quality: Claude Sonnet 4 or GPT-4 Turbo
Throughput: vLLM for production scale

Multi-Provider Setup

Use different providers for different tasks:

typescript
import {
  Agent,
  AnthropicProvider,
  OllamaProvider,
  ToolRegistry,
} from '@lov3kaizen/agentsea-core';

// Cloud provider for complex tasks
const claudeProvider = new AnthropicProvider(process.env.ANTHROPIC_API_KEY);

// Local provider for simple tasks
const ollamaProvider = new OllamaProvider();

const toolRegistry = new ToolRegistry();

// Complex reasoning agent (cloud)
const researchAgent = new Agent(
  {
    name: 'researcher',
    model: 'claude-sonnet-4-20250514',
    provider: 'anthropic',
    systemPrompt: 'You are a research assistant.',
  },
  claudeProvider,
  toolRegistry,
);

// Simple task agent (local, free)
const helperAgent = new Agent(
  {
    name: 'helper',
    model: 'llama3.2',
    provider: 'ollama',
    systemPrompt: 'You are a helpful assistant.',
  },
  ollamaProvider,
  toolRegistry,
);

// Use the right agent for each task
const complexResult = await researchAgent.execute('Analyze this...', context);
const simpleResult = await helperAgent.execute('What is 2+2?', context);

Next Steps

💡 Pro Tip

Start with a cloud provider to validate your idea, then migrate to local providers for cost savings at scale. Use the same AgentSea ADK code - just swap the provider! Many successful products save $75K+ annually by running production workloads on self-hosted models.