Providers & Models

VoltAgent is built directly on top of the Vercel AI SDK. You can either:

• Pass a LanguageModel from an ai-sdk provider package, or
• Use a model string like openai/gpt-4o-mini and let VoltAgent resolve it with the built-in model router.

Both approaches are fully compatible with ai-sdk streaming, tool calling, and structured outputs. For the router, VoltAgent ships with a registry snapshot generated from models.dev.

Model Strings (Model Router)

Model strings remove the need to import provider packages in your app:

import { Agent } from "@voltagent/core";

const agent = new Agent({
  name: "my-agent",
  instructions: "You are a helpful assistant",
  model: "openai/gpt-4o-mini",
});

Other examples:

const claudeAgent = new Agent({
  name: "claude-agent",
  instructions: "Answer with concise reasoning",
  model: "anthropic/claude-3-5-sonnet",
});

const geminiAgent = new Agent({
  name: "gemini-agent",
  instructions: "Respond in Turkish",
  model: "google/gemini-2.0-flash",
});

If you need provider-specific configuration or want to use the ai-sdk APIs directly, pass a LanguageModel instead. See Model Router & Registry for how strings are resolved and how env vars are mapped. For the full provider directory, see Models.

Installation

Install the AI SDK base package (required):

npm

npm install ai

yarn

yarn add ai

pnpm

pnpm add ai

If you plan to import ai-sdk providers directly (for embeddings or provider-specific helpers like openai.embedding(...)), install those packages too. If you only use model strings, you can skip them:

npm

# For example, to use OpenAI:
npm install @ai-sdk/openai

# Or Anthropic:
npm install @ai-sdk/anthropic

# Or Google:
npm install @ai-sdk/google

yarn

# For example, to use OpenAI:
yarn add @ai-sdk/openai

# Or Anthropic:
yarn add @ai-sdk/anthropic

# Or Google:
yarn add @ai-sdk/google

pnpm

# For example, to use OpenAI:
pnpm add @ai-sdk/openai

# Or Anthropic:
pnpm add @ai-sdk/anthropic

# Or Google:
pnpm add @ai-sdk/google

If you only use model strings, you can skip installing provider packages.

Usage Examples

Model Strings

import { Agent } from "@voltagent/core";

const agent = new Agent({
  name: "my-agent",
  instructions: "You are a helpful assistant",
  model: "openai/gpt-4o-mini",
});

Direct ai-sdk Provider

import { Agent } from "@voltagent/core";
import { openai } from "@ai-sdk/openai";

const agent = new Agent({
  name: "my-agent",
  instructions: "You are a helpful assistant",
  model: openai("gpt-4o-mini"),
});

Available Providers

The lists below describe ai-sdk provider packages. Model strings are resolved through VoltAgent's registry and map to these providers under the hood.

First-Party AI SDK Providers

These providers are maintained by Vercel and offer the highest level of support and integration:

Foundation Models

Provider	Package	Documentation	Key Models
xAI Grok	@ai-sdk/xai	Docs	grok-4, grok-3, grok-2-vision
OpenAI	@ai-sdk/openai	Docs	gpt-4.1, gpt-4o, o3, o1
Anthropic	@ai-sdk/anthropic	Docs	claude-opus-4, claude-sonnet-4, claude-3.5
Google Generative AI	@ai-sdk/google	Docs	gemini-2.0-flash, gemini-1.5-pro
Google Vertex	@ai-sdk/google-vertex	Docs	gemini models, claude models via Vertex
Mistral	@ai-sdk/mistral	Docs	mistral-large, pixtral-large, mistral-medium

Cloud Platforms

Provider	Package	Documentation	Description
Amazon Bedrock	@ai-sdk/amazon-bedrock	Docs	Access to various models via AWS
Azure OpenAI	@ai-sdk/azure	Docs	OpenAI models via Azure
Vercel	@ai-sdk/vercel	Docs	v0 model for code generation

Specialized Providers

Provider	Package	Documentation	Specialization
Groq	@ai-sdk/groq	Docs	Ultra-fast inference
Together.ai	@ai-sdk/togetherai	Docs	Open-source models
Cohere	@ai-sdk/cohere	Docs	Enterprise search & generation
Fireworks	@ai-sdk/fireworks	Docs	Fast open-source models
DeepInfra	@ai-sdk/deepinfra	Docs	Affordable inference
DeepSeek	@ai-sdk/deepseek	Docs	DeepSeek models including reasoner
Cerebras	@ai-sdk/cerebras	Docs	Fast Llama models
Perplexity	@ai-sdk/perplexity	Docs	Search-enhanced responses

Audio & Speech Providers

Provider	Package	Documentation	Specialization
ElevenLabs	@ai-sdk/elevenlabs	Docs	Text-to-speech
LMNT	@ai-sdk/lmnt	Docs	Voice synthesis
Hume	@ai-sdk/hume	Docs	Emotional intelligence
Rev.ai	@ai-sdk/revai	Docs	Speech recognition
Deepgram	@ai-sdk/deepgram	Docs	Speech-to-text
Gladia	@ai-sdk/gladia	Docs	Audio intelligence
AssemblyAI	@ai-sdk/assemblyai	Docs	Speech recognition & understanding

Community Providers

These providers are created and maintained by the open-source community:

Provider	Package	Documentation	Description
Ollama	ollama-ai-provider	Docs	Local model execution
FriendliAI	@friendliai/ai-provider	Docs	Optimized inference
Portkey	@portkey-ai/vercel-provider	Docs	LLM gateway & observability
Cloudflare Workers AI	workers-ai-provider	Docs	Edge AI inference
OpenRouter	@openrouter/ai-sdk-provider	Docs	Unified API for multiple providers
Requesty	@requesty/ai-sdk	Docs	Request management
Crosshatch	@crosshatch/ai-provider	Docs	Specialized models
Mixedbread	mixedbread-ai-provider	Docs	Embedding models
Voyage AI	voyage-ai-provider	Docs	Embedding models
Mem0	@mem0/vercel-ai-provider	Docs	Memory-enhanced AI
Letta	@letta-ai/vercel-ai-sdk-provider	Docs	Stateful agents
Spark	spark-ai-provider	Docs	Chinese language models
AnthropicVertex	anthropic-vertex-ai	Docs	Claude via Vertex AI
LangDB	@langdb/vercel-provider	Docs	Database-aware AI
Dify	dify-ai-provider	Docs	LLMOps platform
Sarvam	sarvam-ai-provider	Docs	Indian language models
Claude Code	ai-sdk-provider-claude-code	Docs	Code-optimized Claude
Built-in AI	built-in-ai	Docs	Browser-native AI
Gemini CLI	ai-sdk-provider-gemini-cli	Docs	CLI-based Gemini
A2A	a2a-ai-provider	Docs	Specialized models
SAP-AI	@mymediset/sap-ai-provider	Docs	SAP AI Core integration

OpenAI-Compatible Providers

For providers that follow the OpenAI API specification:

Provider	Documentation	Description
LM Studio	Docs	Local model execution with GUI
Baseten	Docs	Model deployment platform
Any OpenAI-compatible API	Docs	Custom endpoints

Model Capabilities

The AI providers support different language models with various capabilities. Here are the capabilities of popular models:

Provider	Model	Image Input	Object Generation	Tool Usage	Tool Streaming
xAI Grok	grok-4	❌	✅	✅	✅
xAI Grok	grok-3	❌	✅	✅	✅
xAI Grok	grok-3-fast	❌	✅	✅	✅
xAI Grok	grok-3-mini	❌	✅	✅	✅
xAI Grok	grok-3-mini-fast	❌	✅	✅	✅
xAI Grok	grok-2-1212	❌	✅	✅	✅
xAI Grok	grok-2-vision-1212	✅	✅	✅	✅
xAI Grok	grok-beta	❌	✅	✅	✅
xAI Grok	grok-vision-beta	✅	❌	❌	❌
Vercel	v0-1.0-md	✅	✅	✅	✅
OpenAI	gpt-4.1	✅	✅	✅	✅
OpenAI	gpt-4.1-mini	✅	✅	✅	✅
OpenAI	gpt-4.1-nano	✅	✅	✅	✅
OpenAI	gpt-4o	✅	✅	✅	✅
OpenAI	gpt-4o-mini	✅	✅	✅	✅
OpenAI	gpt-4	❌	✅	✅	✅
OpenAI	o3-mini	❌	❌	✅	✅
OpenAI	o3	❌	❌	✅	✅
OpenAI	o4-mini	❌	❌	✅	✅
OpenAI	o1	✅	❌	✅	✅
OpenAI	o1-mini	✅	❌	✅	✅
OpenAI	o1-preview	❌	❌	❌	❌
Anthropic	claude-opus-4-20250514	✅	✅	✅	✅
Anthropic	claude-sonnet-4-20250514	✅	✅	✅	✅
Anthropic	claude-3-7-sonnet-20250219	✅	✅	✅	✅
Anthropic	claude-3-5-sonnet-20241022	✅	✅	✅	✅
Anthropic	claude-3-5-sonnet-20240620	✅	✅	✅	✅
Anthropic	claude-3-5-haiku-20241022	✅	✅	✅	✅
Mistral	pixtral-large-latest	✅	✅	✅	✅
Mistral	mistral-large-latest	❌	✅	✅	✅
Mistral	mistral-medium-latest	❌	✅	✅	✅
Mistral	mistral-medium-2505	❌	✅	✅	✅
Mistral	mistral-small-latest	❌	✅	✅	✅
Mistral	pixtral-12b-2409	✅	✅	✅	✅
Google Generative AI	gemini-2.0-flash-exp	✅	✅	✅	✅
Google Generative AI	gemini-1.5-flash	✅	✅	✅	✅
Google Generative AI	gemini-1.5-pro	✅	✅	✅	✅
Google Vertex	gemini-2.0-flash-exp	✅	✅	✅	✅
Google Vertex	gemini-1.5-flash	✅	✅	✅	✅
Google Vertex	gemini-1.5-pro	✅	✅	✅	✅
DeepSeek	deepseek-chat	❌	✅	✅	✅
DeepSeek	deepseek-reasoner	❌	❌	❌	❌
Cerebras	llama3.1-8b	❌	✅	✅	✅
Cerebras	llama3.1-70b	❌	✅	✅	✅
Cerebras	llama3.3-70b	❌	✅	✅	✅
Groq	meta-llama/llama-4-scout-17b-16e-instruct	✅	✅	✅	✅
Groq	llama-3.3-70b-versatile	❌	✅	✅	✅
Groq	llama-3.1-8b-instant	❌	✅	✅	✅
Groq	mixtral-8x7b-32768	❌	✅	✅	✅
Groq	gemma2-9b-it	❌	✅	✅	✅

Note: This table is not exhaustive. Additional models can be found in the provider documentation pages and on the provider websites.

Migration from Deprecated Providers

If you're currently using VoltAgent's native providers (@voltagent/anthropic-ai, @voltagent/google-ai, @voltagent/groq-ai), we recommend migrating to the Vercel AI SDK providers:

Before (Deprecated):

import { AnthropicProvider } from "@voltagent/anthropic-ai";

const provider = new AnthropicProvider({ apiKey: "..." });
const agent = new Agent({
  llm: provider,
  model: "claude-opus-4-1",
});

After (Recommended):

import { Agent } from "@voltagent/core";

const agent = new Agent({
  model: "anthropic/claude-3-5-sonnet",
  instructions: "You are a helpful assistant",
});

Environment Variables

Most providers use environment variables for API keys:

# OpenAI
OPENAI_API_KEY=your-key

# Anthropic
ANTHROPIC_API_KEY=your-key

# Google
GOOGLE_GENERATIVE_AI_API_KEY=your-key

# Groq
GROQ_API_KEY=your-key

# And so on...

OpenAI-compatible providers may also need a base URL. You can set a provider-specific override like:

OPENROUTER_BASE_URL=https://openrouter.ai/api/v1

Next Steps

1. Choose a provider based on your needs (performance, cost, capabilities)
2. Install the corresponding package if you plan to import the provider
3. Configure your API keys
4. Start building with VoltAgent!

For detailed information about each provider, visit the Vercel AI SDK documentation.

---

Acknowledgments

The provider lists and model capabilities in this documentation are sourced from the Vercel AI SDK documentation.

A special thanks to the Vercel AI SDK maintainers and community for creating and maintaining this comprehensive ecosystem of AI providers. Their work enables developers to seamlessly integrate with 30+ AI providers through a unified, well-designed interface.

VoltAgent builds upon this excellent foundation to provide a complete framework for building AI agents and workflows.