Ecosystem

Last updated January 21, 2026

AI Gateway integrates with the AI development ecosystem you use. Whether you're building with LangChain, LlamaIndex, or other popular frameworks, connect through compatible APIs and get unified billing, observability, and model access.

Framework integrations

These popular frameworks work through OpenAI-compatible endpoints or native integrations:

Framework	Language	Integration type	Use case
LangChain	Python/JS	OpenAI-compatible	Chains, agents, RAG pipelines
LlamaIndex	Python	Native package	Knowledge assistants, document Q&A
Mastra	TypeScript	Native	AI workflows and agents
Pydantic AI	Python	Native	Type-safe agents, structured outputs
LiteLLM	Python	Native prefix	Unified LLM interface
Langfuse	Any	Observability	LLM analytics and tracing

LangChain

Connect LangChain through the OpenAI-compatible endpoint:

from langchain_openai import ChatOpenAI
 
llm = ChatOpenAI(
    model="anthropic/claude-sonnet-4.5",
    api_key=os.getenv("AI_GATEWAY_API_KEY"),
    base_url="https://ai-gateway.vercel.sh/v1"
)
 
response = llm.invoke("Explain RAG in one sentence")

LlamaIndex

Use the dedicated llama-index-llms-vercel-ai-gateway package:

pip install llama-index-llms-vercel-ai-gateway

from llama_index.llms.vercel_ai_gateway import VercelAIGateway
 
llm = VercelAIGateway(
    model="anthropic/claude-sonnet-4.5",
    api_key=os.getenv("AI_GATEWAY_API_KEY")
)

Pydantic AI

Pydantic AI has a native VercelProvider for type-safe AI agents:

from pydantic_ai import Agent
from pydantic_ai.providers.vercel import VercelProvider
 
agent = Agent(
    VercelProvider(model="anthropic/claude-sonnet-4.5"),
    system_prompt="You are a helpful assistant"
)
 
result = agent.run_sync("What is the capital of France?")

See the Framework Integrations documentation for complete setup guides.

App attribution

App Attribution lets you identify your application in requests. When you include attribution headers, Vercel can feature your app—increasing visibility for your project.

Add attribution to your requests:

const response = await fetch('https://ai-gateway.vercel.sh/v1/chat/completions', {
  headers: {
    'Authorization': `Bearer ${apiKey}`,
    'X-Vercel-AI-App-Name': 'My AI App',
    'X-Vercel-AI-App-Url': 'https://myaiapp.com',
  },
  // ... request body
});