AgentSea - Unite and Orchestrate AI Agents

🔌 12+ Providers Supported

Cloud Providers (6): Anthropic, OpenAI, Google Gemini, Mistral AI, DeepSeek, xAI (Grok)

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 (selection):
// - claude-opus-4-6 (Most capable)
// - claude-sonnet-4-5-20250929 (Balanced, latest Sonnet)
// - claude-haiku-4-5 (Fastest, cheapest)
// - claude-3-7-sonnet-20250219 (Extended thinking)

const agent = new Agent(
  {
    model: 'claude-sonnet-4-5-20250929',
    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 the GPT and o-series and a strong ecosystem. AgentSea ships with a registry of 60+ models across providers, including the latest OpenAI releases.

typescript

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

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

// Available models (selection):
// - gpt-5 (Most capable)
// - gpt-4.1 (Strong general-purpose)
// - gpt-4o (Multimodal, fast)
// - gpt-4o-mini (Fast, cost-effective)
// - o3 (Advanced reasoning)
// - o4-mini (Efficient reasoning)

const agent = new Agent(
  {
    model: 'gpt-4.1',
    provider: 'openai',
    temperature: 0.8,
    maxTokens: 2048,
  },
  provider,
  toolRegistry,
);

Reasoning models such as o3 and o4-mini have model-specific options. Use per-model type safety to get compile-time validation of which options each model supports.

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 { GeminiProvider } from '@lov3kaizen/agentsea-core';

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

// Available models:
// - gemini-2.5-pro (Most capable)
// - gemini-2.5-flash (Fast, affordable)
// - gemini-2.0-flash (Low latency)

const agent = new Agent(
  {
    model: 'gemini-2.5-flash',
    provider: 'gemini',
    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

Mistral, DeepSeek & xAI

Mistral AI, DeepSeek, and xAI (Grok) are first-class providers in the Provider registry. They are created with the same type-safe config builders as the other providers and run over an OpenAI-compatible transport under the hood — just set the matching API key environment variable.

typescript

import {
  createProvider,
  mistral,
  deepseek,
  xai,
} from '@lov3kaizen/agentsea-core';

// Mistral AI — set MISTRAL_API_KEY
const mistralProvider = createProvider(
  mistral('mistral-large-latest', { tools: [myTool] }),
);

// DeepSeek — set DEEPSEEK_API_KEY
const deepseekProvider = createProvider(deepseek('deepseek-reasoner'));

// xAI (Grok) — set XAI_API_KEY
const grokProvider = createProvider(xai('grok-3'));

Mistral AI

European AI with strong open models and competitive pricing.

mistral-large-latest, mistral-small-latest, codestral-latest

DeepSeek

Cost-efficient chat and reasoning models with strong coding ability.

deepseek-chat, deepseek-reasoner

xAI (Grok)

Grok models with real-time knowledge and fast inference tiers.

grok-3, grok-3-fast, grok-3-mini

All providers share the same config-builder API (anthropic(), openai(), gemini(), mistral(), deepseek(), xai()) so you get compile-time validation of which options each model supports. See per-model type safety.

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

Provider	Quality	Speed	Cost	Context	Best For
Anthropic	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	$$$	200K	Reasoning, safety
OpenAI	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	$$$	128K	General purpose
Google Gemini	⭐⭐⭐⭐⭐	⭐⭐⭐⭐⭐	$$	1M+	Multimodal, long context
Mistral AI	⭐⭐⭐⭐	⭐⭐⭐⭐	$$	128K	European compliance
DeepSeek	⭐⭐⭐⭐	⭐⭐⭐⭐	$	128K	Reasoning, coding (low cost)
xAI (Grok)	⭐⭐⭐⭐	⭐⭐⭐⭐	$$	128K	Real-time knowledge

Local Providers

Provider	Ease of Use	Performance	Features	Best For
Ollama	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	Model mgmt, CLI	Getting started
LM Studio	⭐⭐⭐⭐⭐	⭐⭐⭐⭐	GUI, easy setup	Non-technical users
LocalAI	⭐⭐⭐	⭐⭐⭐⭐	Multi-modal, Docker	Self-hosted services
vLLM	⭐⭐	⭐⭐⭐⭐⭐	PagedAttention	Production scale
Text Gen WebUI	⭐⭐⭐⭐	⭐⭐⭐	Web UI, extensions	Experimentation
Jan	⭐⭐⭐⭐⭐	⭐⭐⭐	Desktop app	ChatGPT 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 Opus 4.6 or GPT-5
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

Voice Providers - STT and TTS providers
Local Models Guide - Deep dive into local execution
CLI Tool - Manage providers with CLI
Agent Configuration - Configure agents with providers
View Examples - Provider usage examples

💡 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.

Provider Reference

🔌 12+ Providers Supported

Provider Categories

Cloud Providers

Local Providers

Voice Providers

Cloud Providers

Anthropic (Claude)

OpenAI (GPT)

Google (Gemini)

Mistral, DeepSeek & xAI

Mistral AI

DeepSeek

xAI (Grok)

Local Providers

🔒 Why Local?

Ollama (Recommended)

LM Studio

LocalAI

Text Generation WebUI

vLLM

Jan

Provider Comparison

Cloud Providers

Local Providers

Choosing a Provider

🎯 For Getting Started

💰 For Cost Savings

🔒 For Privacy

⚡ For Performance

Multi-Provider Setup

Next Steps

💡 Pro Tip