AI Coding Agent Platforms
Mastra pricing: what’s free today, what’s “free to start,” and what happens when Platform pricing launches in Q1 2026?
7 min read
Most teams discover Mastra through GitHub or npm create mastra long before they ever talk to our sales team, so the obvious question is: what exactly is free, what stays free, and what happens when Mastra Platform pricing goes live in Q1 2026? This FAQ breaks down how Mastra’s open-source framework, Mastra Cloud, and future Platform pricing fit together so you can build confidently today and avoid surprise bills later.
Quick Answer: The Mastra framework is and will remain free, open-source, and Apache 2.0–licensed. Mastra Cloud and “Mastra Platform” are “free to start” with generous usage included for small projects, and we’ll introduce paid usage-based plans in Q1 2026 that you can opt into once you outgrow the free tier.
Frequently Asked Questions
What’s free in Mastra today?
Short Answer: The core Mastra framework is fully free and open-source forever; Mastra Cloud is currently in a “free to start” phase with usage-based limits we’ll formalize as Platform pricing in Q1 2026.
Expanded Explanation:
Mastra’s core is an Apache 2.0 open-source TypeScript framework. You can run Mastra entirely on your own infrastructure—Next.js, Express, Hono, or any Node environment—without paying Mastra a cent. That includes core primitives like Agent, Workflow, RAG, Memory, MCP (client and server), and local Observability.
Mastra Cloud and the emerging “Mastra Platform” are cloud services that sit alongside the OSS framework: hosted Studio, managed observability, and (over time) things like multi-project management, secure sharing, and team features. Today, those services are effectively free for early adopters with reasonable usage. When Platform pricing launches in Q1 2026, those same services will move to a clear free tier plus paid plans.
Key Takeaways:
- The Mastra framework is Apache 2.0 and free to use in dev and production, self-hosted.
- Cloud/Platform features are “free to start” now and will get formal free + paid tiers in Q1 2026.
How does “free to start” work for Mastra Cloud and Platform?
Short Answer: “Free to start” means you can build and ship real workloads on Mastra Cloud with generous limits before any billing kicks in; once Platform pricing is live, those limits will simply map to a defined free tier.
Expanded Explanation:
From a builder’s perspective, you should be able to go from npm create mastra → shipping an agent into production without negotiating a contract. That’s what “free to start” is for. You can connect your own model keys, run agents in production, and send traces to Mastra Cloud to debug and evaluate—without an immediate paywall.
Behind the scenes, Mastra Cloud tracks usage (traces, projects, storage, etc.) so that when Mastra Platform’s pricing launches in Q1 2026, your current usage naturally maps onto a free or paid tier. If you’re under the free tier limits, nothing changes. If you’re over, you’ll have a clear path to a paid plan instead of surprise lockouts.
Steps:
- Start with OSS: Install and run Mastra purely from open-source packages in your own stack.
- Optionally connect Cloud: Enable Mastra Cloud observability/exporters for better traces and Studio experience.
- Watch for Q1 2026: When Platform pricing goes live, align your current usage with the published free and paid tiers and decide whether to stay on free or upgrade.
What’s the difference between “free forever” OSS and paid Platform?
Short Answer: The framework (TypeScript packages you install) is free forever; the Platform is a hosted layer (Cloud, Studio, team features, managed observability) that becomes usage-priced in Q1 2026.
Expanded Explanation:
Think of Mastra as two distinct but complementary layers:
-
The open-source framework is your runtime:
Agent,Workflow, RAG modules, MCPClient/MCPServer, Memory, Evals, Observability primitives, and exporters you run where your app lives. This layer is Apache 2.0, under your control, and deployable anywhere you host Node/TypeScript. It’s how Plaid, Elastic, Replit, Docker, and SoftBank integrate Mastra into existing infrastructure. -
The Platform is your control plane: hosted Studio UI, centralized traces, multi-project views, team access, and future capabilities (policy, governance, advanced eval dashboards, etc.). This is what will have clear free and paid tiers. It doesn’t replace self-hosting; it gives you a managed layer around it.
Comparison Snapshot:
- Option A: OSS-only (Free Forever):
Self-hosted agents/workflows, local or self-managed observability, your infra, no Mastra bill. - Option B: OSS + Platform (Free to Start → Paid):
Same OSS core, plus Mastra Cloud for tracing, Studio, team features, and managed components once limits are exceeded. - Best for:
- OSS-only: Tight compliance, fully in-house infra, teams with existing observability stacks.
- OSS + Platform: Fast iteration, teams wanting hosted Studio, managed traces, and fewer infra decisions.
What happens to existing users when Platform pricing launches in Q1 2026?
Short Answer: Existing users keep running; you won’t lose agents or data overnight. You’ll see clear usage dashboards, get mapped to a default tier, and then choose whether to stay on free limits or upgrade to a paid plan.
Expanded Explanation:
Pricing changes are scary when your agents are in production; we treat that as a reliability issue, not a marketing moment. When Mastra Platform pricing goes live:
- Your current Mastra Cloud projects and traces will remain accessible.
- You’ll see which tier your usage maps to—ideally before any billing is enforced.
- We’ll give you time and tooling to adjust: trimming old traces, reducing retention, or switching exporters (e.g., to OpenTelemetry and your own ClickHouse cluster) if you want to minimize Platform usage.
The goal is to make pricing a configuration decision, not a fire drill. Your agents will keep working, because the runtime is in your codebase—not gated by a remote control plane.
What You Need:
- Awareness of usage: Understand how many traces, projects, and how much storage you’re sending to Mastra Cloud before Q1 2026.
- A migration/upgrade plan: Decide whether you’ll stay fully in free limits, upgrade to a paid Platform plan, or move heavy observability to your own OpenTelemetry-compatible backend.
How should teams plan strategically for Mastra pricing over the next 12–24 months?
Short Answer: Treat Mastra’s OSS framework as a stable, free runtime, and design your observability and Cloud usage so you can flex between a free Platform tier, a paid plan, or your own infra without rewriting agents.
Expanded Explanation:
From a systems perspective, the risk isn’t paying for a Platform—it’s getting locked into a control plane you can’t outgrow. Mastra’s design is intentionally code-first and exporter-based so that you can:
- Start with Mastra Cloud for speed and Studio UX.
- Gradually externalize heavy observability to your own storage when volume and cost demand it.
- Keep agents, workflows, MCP tools, and evals defined in TypeScript in your repo, independent of Platform pricing.
Practically, that means treating Mastra Platform like you treat any cloud database or log service: great when it accelerates you, replaceable when you need specialized infra or cost controls.
Why It Matters:
- Cost predictability: You can scale from hobby projects to Plaid/Elastic/Replit scale by tuning exporters and storage backends instead of rewriting agents.
- Operational control: Because agents/workflows live in your app, on your infra, you retain control over reliability, performance, and compliance, regardless of how you use Platform features.
Quick Recap
The Mastra framework—agents, workflows, RAG, memory, MCP, evals, observability primitives—is Apache 2.0 and free to use, now and in the future. Mastra Cloud and the broader Mastra Platform are “free to start” today, offering hosted Studio and managed observability, with usage-based pricing coming in Q1 2026. When that happens, you’ll keep running as-is, get mapped to clear free and paid tiers, and have full flexibility to stay on free limits, upgrade, or shift heavy workloads to your own infra.