Mastra
Also known as: Mastra Studio, Mastra Platform
TypeScript agent framework with evals, workflow orchestration, observability, and Mastra Studio control plane, with dual licensing.
Mastra is an open-source, TypeScript-native framework for building AI agents and AI-powered applications. It exists to close a gap that frustrated many web developers: most serious agent tooling grew up in the Python ecosystem, leaving JavaScript and TypeScript teams to stitch together thin wrappers and a dozen unrelated packages. Built by the team behind the Gatsby web framework, Mastra is designed from the ground up for the language millions of web developers already use, with the primitives for building agents bundled into one coherent stack.
The framework provides agents, tools, memory, workflows, and evaluation as first-class building blocks. An agent is defined in one place with its instructions, model, tools, and memory, and then reasons about a goal, decides which tools to call, and iterates until it reaches an answer. Tools are typed and validated and can be shared across agents through the Model Context Protocol, with optional human approval required before a tool actually runs. Memory persists conversation context and application state across sessions, ranging from simple message history to memory that learns about users over time.
For work that needs structure rather than open-ended reasoning, Mastra's workflows let teams orchestrate multi-step processes with type-safe control flow: sequential steps, parallel branches, conditions, and loops, mixing agent reasoning with deterministic code. A workflow can suspend, hand control to a human or external system, and resume when input arrives, with its state persisted automatically, which suits durable, long-running processes.
Mastra also ships the surrounding production tooling: a unified model router that reaches thousands of models across dozens of providers through one interface, guardrails that sanitize inputs and outputs to resist prompt injection, scorers and evals to check agent quality against repeatable benchmarks before changes ship, and built-in tracing of decisions, tokens, latency, and cost. A visual studio lets teams test agents, follow execution traces, and tune prompts without code. Applications deploy to any Node.js environment, with one-click deployers for common platforms and enterprise options including role-based access and keeping data inside a private network. It is used by engineering teams building production agents on a modern JavaScript stack.
Vendor details
Canonical URL
https://mastra.ai/
Category
Agent infrastructure
Company status
independent
Use cases & customers
Target customers
Deployment options
In practice
Your team works in TypeScript but every serious agent framework assumes Python. Mastra is TypeScript-native, giving you agents, tools, memory, and workflows in one stack that integrates cleanly with Next.js and Node.
Part of your process needs an agent's judgment and part needs deterministic steps in a fixed order. Mastra's workflows mix agent reasoning with plain code in type-safe control flow, and can pause for human approval before resuming.
You don't want to commit to one model provider or wire up evals and tracing yourself. Mastra routes across thousands of models through one interface and ships built-in scorers, guardrails, and observability to make agents production-ready.
Agentic Index coverage score
11.0 / 14 capabilities · 79%
| Integrations & Tool CallingAgentic AI Index Expansion Evidence Report | Full |
|---|---|
| Workflow OrchestrationAgentic AI Index Expansion Evidence Report | Full |
| Knowledge Grounding & RAGAgentic AI Index Expansion Evidence Report | Unable to verify |
| Human Oversight & GuardrailsAgentic AI Index Expansion Evidence Report | Unable to verify |
| Security, Identity & GovernanceAgentic AI Index Expansion Evidence Report | Full |
| Observability & AuditabilityAgentic AI Index Expansion Evidence Report | Full |
| Memory & State PersistenceAgentic AI Index Expansion Evidence Report | Full |
| Deployment & Data ResidencyAgentic AI Index Expansion Evidence Report | Full |
| Prebuilt Agents, Templates & PacksAgentic AI Index Expansion Evidence Report | Full |
| Triggers & Channel CoverageAgentic AI Index Expansion Evidence Report | Partial |
| Model Flexibility & RoutingAgentic AI Index Expansion Evidence Report | Partial |
| APIs, SDKs & MCP ExtensibilityAgentic AI Index Expansion Evidence Report | Full |
| Testing, Debugging & OptimizationAgentic AI Index Expansion Evidence Report | Full |
| Browser & Computer UseAgentic AI Index Expansion Evidence Report | Full |
The Agentic Index coverage score grades every vendor Full, Partial or Unable to verify against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Recent platform changes
Mastra released version 1.56.0 of its core framework, introducing persistable declarative workflows. Developers can now author workflows as JSON data graphs using new declarative step entry types like agent, tool, and mapping. This update includes end-to-end support for stored workflows via HTTP APIs, a client SDK, and database persistence.
Bears on: Workflow orchestration
View sourceMastra launched several updates to its hosting platform, including deployment environments for isolating production and staging workloads, EU and US geographic regions for localized data hosting, and managed workspaces that equip agents with persistent filesystems and secure sandboxing.
Bears on: Deployment / data residency
View sourceMastra released Memory Extractors, a feature that enables agents to automatically pull and persist structured data from natural language conversations.
Bears on: Memory / state
View sourcePricing
From $250/mo · free tier
hybrid
Included quota
OSS framework (free, Apache 2.0): agents, workflows (suspend/resume, human-in-the-loop), RAG, memory, tool selection, evals, local playground with tracing, and the model router. Platform Starter ($0): hosted Studio + deployments + observability to get started (unlimited users/deployments). Teams ($250/team/mo): scaled team usage of the Platform, metered on observability events/CPU/egress. Enterprise: on-prem deployment, custom SLAs, production-Studio RBAC, and dedicated Slack support.
What is public
Mastra (mastra.ai - open-source TypeScript framework for production AI agents + the hosted Mastra Platform) is free as OSS under Apache 2.0 (self-host anywhere on Node.js; enterprise features in ee/ dirs are source-available under a separate Mastra Enterprise License). The Mastra Platform (launched April 9, 2026) has three tiers: Starter $0/mo (free, unlimited users/deployments), Teams $250/team/mo (metered on observability events, CPU time, and data egress; Memory Gateway metered on memory tokens + retrieval storage), and custom Enterprise (on-prem, custom SLAs, dedicated Slack support, production-Studio RBAC). You bring your own model API keys, so LLM usage is paid separately to providers.
Billing mechanics
Two paths: (1) self-host the Apache-2.0 framework free (deploy to Vercel, Cloudflare, Netlify, or any Node.js runtime - you cover your own infra + LLM provider costs); or (2) the managed Mastra Platform with per-team tiers (Starter $0, Teams $250/team/mo, Enterprise custom) - the Teams tier is metered on observability events, CPU time, and data egress (Memory Gateway: memory tokens + retrieval storage). The Platform adds cloud Studio, GitHub-connected deployments with autoscaling + instant rollbacks, and centralized observability (structured logs, AI-aware tracing, eval dashboards). Model usage is always pass-through - Mastra's model router spans 40+ providers via your own keys.
Cost watchouts
LLM provider costs are separate (bring your own keys) and scale with agent runs; self-hosting shifts infra/ops cost to you; the Platform is newly launched (April 2026) so tier inclusions/limits are still settling; TypeScript/JavaScript only - no Python support by design
Variable cost rationale
Platform tiers are flat per-team monthly but metered on observability events/CPU/egress; real cost is dominated by LLM provider usage (pass-through via your own keys) that scales with agent runs; self-hosting trades the Platform fee for your own variable infra + model spend
Additional watchouts
TypeScript/JavaScript only (no Python) - a deliberate scope choice that excludes Python-first teams; Platform is brand-new (limits/pricing still maturing); you own LLM provider costs and (if self-hosting) all infra/ops; agent cost-per-run visibility requires the Platform's observability
Overage / add-ons
Self-hosting has no Mastra-imposed caps (you pay your own infra + model usage). On the Platform, scaling beyond the Starter/Teams tiers moves you up a tier (Teams -> Enterprise); the Teams tier meters observability events, CPU time, and data egress; LLM token usage is always billed by the model providers via your own keys, regardless of tier.
Sales call required
Mixed (some tiers require a call)
Free / trial
Free tier
Lowest paid plan
Teams $250/mo
Commercial notes
Built by Gatsby veterans (Sam Bhagwat/Shane Thomas/Abhi Aiyer), YC W25, $13M seed (Oct 2025, 120+ investors incl. Gradient Ventures); Mastra 1.0 stable Jan 2026; 300,000+ weekly NPM downloads, 21,800+ GitHub stars (one of the fastest-growing JS frameworks); production users include Replit (generates thousands of Mastra agents daily), PayPal, Adobe, SoftBank, Marsh McLennan (75,000+ employees), Brex, Sanity; built on Vercel's AI SDK
Key ambiguities
The Platform launched April 9, 2026; the Teams tier ($250/team/mo) is metered on observability events/CPU/data egress (Studio/Server) and memory tokens/retrieval storage (Memory Gateway), but exact included quotas/overage rates beyond the headline prices ($0 / $250 / custom) aren't fully published yet
Cancellation / refund
OSS is free forever (Apache 2.0, no contract); Platform Starter is free and Teams is self-serve monthly ($250/team); Enterprise is custom-contracted (on-prem, SLAs) via sales
Support SLA / resale
Community support (OSS); Enterprise adds custom SLAs and dedicated Slack support; deploys to Vercel/Cloudflare/Netlify/any Node.js host or Mastra Platform; MCP integration; model router with automatic fallbacks across 40+ providers; vector DB support (Pinecone, Qdrant, pgvector, Astra, etc.)
Missing data
Platform per-tier included quotas/overage rates beyond the $0/$250/custom headline aren't fully public yet (launched April 2026; Teams is metered on observability events/CPU/egress + memory tokens/storage). Seed 'Free tier; Teams $250/mo; OSS self-hosted free' is accurate and current - the core OSS license is Apache 2.0 (enterprise ee/ features are source-available under the Mastra Enterprise License).
Related vendors
- Acrab — Singapore compute infrastructure company building a full stack…
- AgentOps — Agent observability and reliability platform with broad model and…
- Agno — High-performance agent runtime and framework (formerly Phidata) with…
- AIsa — Unified resource and payment gateway for AI agents that lets them…
- AlphaBitCore — AI control plane that governs how models, agents, tools, and…
- Anchor Browser — Cloud hosted browser infrastructure that lets AI agents operate real…
Alternatives to Mastra
The closest documented capability profiles to Mastra among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Browserless9.5 / 14Adds documented Human Oversight & Guardrails
- Pydantic AI12.5 / 14Adds documented Knowledge Grounding & RAG and Human Oversight & Guardrails
- SmythOS10.5 / 14Adds documented Knowledge Grounding & RAG
- Anchor Browser8.0 / 14A lighter documented profile than Mastra
- Apify9.0 / 14Adds documented Knowledge Grounding & RAG
- DBOS9.0 / 14Adds documented Human Oversight & Guardrails
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded