Model Providers

Model providers are used to interact with different language models.

Vercel’s AI SDK is one of the most popular Typescript AI libraries. The AI SDK offers model routing (a unified interface on top of OpenAI, Anthropic, etc), structured output, and tool calling.

Mastra is compatible with the AI SDK to help teams build their proof-of-concepts into production-ready apps. When creating agents in Mastra, you can use any AI SDK-compatible model provider.

Here’s a minimal example of an agent using the OpenAI provider from the AI SDK:

src/mastra/agents/weather-agent.ts


import { openai } from "@ai-sdk/openai";
import { Agent } from "@mastra/core/agent";
 
const agent = new Agent({
  name: "WeatherAgent",
  instructions: "Instructions for the agent...",
  model: openai("gpt-4-turbo"),
});
 
const result = await agent.generate("What is the weather like?");

💡

Mastra is built on top of the AI SDK and uses AI SDK-compatible model providers for model routing.

Types of AI SDK model providers

Model providers from the AI SDK can be grouped into three main categories:

Official providers maintained by the AI SDK team
OpenAI-compatible providers
Community providers

You can find a list of all available model providers in the AI SDK documentation .

💡

AI SDK model providers are packages that need to be installed in your Mastra project. The default model provider selected during the installation process is installed in the project.

If you want to use a different model provider, you need to install it in your project as well.

Here are some examples of how Mastra agents can be configured to use the different types of model providers:

Official providers

Official model providers are maintained by the AI SDK team. Their packages are usually prefixed with @ai-sdk/, e.g. @ai-sdk/anthropic, @ai-sdk/openai, etc.

src/mastra/agents/weather-agent.ts


import { openai } from "@ai-sdk/openai";
import { Agent } from "@mastra/core/agent";
 
const agent = new Agent({
  name: "WeatherAgent",
  instructions: "Instructions for the agent...",
  model: openai("gpt-4-turbo"),
});

Additional configuration may be done by importing a helper function from the AI SDK provider. Here’s an example using the OpenAI provider:

src/mastra/agents/weather-agent.ts


import { createOpenAI } from "@ai-sdk/openai";
import { Agent } from "@mastra/core/agent"
 
const openai = createOpenAI({
    baseUrl: "<your-custom-base-url>",
    apiKey: "<your-custom-api-key>",
    ...otherOptions
});
 
const agent = new Agent({
    name: "WeatherAgent",
    instructions: "Instructions for the agent...",
    model: openai("<model-name>"),
});

Different AI providers may have different options for configuration. Please refer to the AI SDK documentation for more information.

OpenAI compatible providers

OpenAI-compatible providers

Some language model providers implement the OpenAI API. For these providers, you can use the @ai-sdk/openai-compatible provider.

Here’s the general setup and provider instance creation:

src/mastra/agents/weather-agent.ts


import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
import { Agent } from "@mastra/core/agent";
 
const openaiCompatible = createOpenAICompatible({
    name: "<model-name>",
    baseUrl: "<base-url>",
    apiKey: "<api-key>",
    headers: {},
    queryParams: {},
    fetch: async (url, options) => {
        // custom fetch logic
        return fetch(url, options);
    }
});
 
const agent = new Agent({
    name: "WeatherAgent",
    instructions: "Instructions for the agent...",
    model: openaiCompatible("<model-name>"),
});

For more information on the OpenAI-compatible provider, please refer to the AI SDK documentation .

Community providers

The AI SDK provides a Language Model Specification . Following this specification, you can create your own model provider compatible with the AI SDK.

Some community providers have implemented this specification and are compatible with the AI SDK. We will look at one such provider, the Ollama provider available in the ollama-ai-provider package.

Here’s an example:

src/mastra/agents/weather-agent.ts


import { ollama } from "ollama-ai-provider";
import { Agent } from "@mastra/core/agent";
 
const agent = new Agent({
    name: "WeatherAgent",
    instructions: "Instructions for the agent...",
    model: ollama("llama3.2:latest"),
});

You can also configure the Ollama provider like so:

src/mastra/agents/weather-agent.ts


import { createOllama } from "ollama-ai-provider";
import { Agent } from "@mastra/core/agent";
 
const ollama = createOllama({
    baseUrl: "<your-custom-base-url>",
    ...otherOptions,
});
 
const agent = new Agent({
    name: "WeatherAgent",
    instructions: "Instructions for the agent...",
    model: ollama("llama3.2:latest"),
});

For more information on the Ollama provider and other available community providers, please refer to the AI SDK documentation .

💡

While this example shows how to use the Ollama provider, other providers like openrouter, azure, etc. may also be used.