Mastra: TypeScript AI Framework

Overview

Mastra is a TypeScript-native AI framework for building agents, workflows, and RAG pipelines. It is built by the team behind Gatsby and raised $36M by early 2026. The core library is open-source; Mastra Cloud is the hosted deployment platform.

Pricing

The open-source framework is free (MIT). Mastra Cloud launched with a "free to start" model; commercial pricing was announced as "launching Q1 2026" as of March 2026. Enterprise features (SSO, RBAC, access control) require a commercial license for production use, but work without a license in development. On-premise deployment and dedicated support are available via an enterprise sales track.

Core primitives

Mastra ships four primitives as first-class concepts:

Agents: LLM-backed agents that reason over goals and invoke tools iteratively
Workflows: Deterministic graphs of steps for complex multi-stage operations
RAG: Retrieval-augmented generation with native connectors to Pinecone, Qdrant, and other stores
Evals: Evaluation pipelines for measuring agent output quality

The cloud platform adds GitHub-connected deployments with autoscaling, structured logs, and eval dashboards.

Model access

Mastra routes to thousands of models across ~94 providers through a single API surface. The models index (as of March 2026) lists 3,300+ models. This breadth is the direct competitor angle against LangChain: same coverage, TypeScript-first, and a more consistent modern API surface.

Where it fits

Mastra is the choice for teams that:

Work primarily in TypeScript / Next.js
Want a modern, opinionated framework (fewer escape hatches than raw LangChain, more than just CrewAI's role-based pattern)
Plan to deploy on Mastra Cloud for managed autoscaling and observability

The trade-off against LangChain: smaller community, less documentation, earlier in maturity. The trade-off against CrewAI: Mastra handles more of the deployment pipeline; CrewAI focuses on the agent orchestration pattern itself.

Field notes

Mastra GitHub repository crossed 12K stars by May 2026 per starhistory.com. Growth is concentrated in the TypeScript/Next.js developer community rather than the Python-heavy LangChain base. [community-thread, 2026-05-01]