Simple, scalable pricing
Pay for what you ship, not for seats. Our AI agent platform pricing scales from your first free experiment to production fleets of autonomous agents — with transparent run-based metering and no surprise bills.
- Free plan
- 14-day Pro trial
- No credit card
Choosing an AI agent platform shouldn't mean decoding a confusing pricing matrix. We bill on a single, honest unit — the agent run — so your costs track the value your agents create, not the number of people who log in. Every plan includes the full agentic stack: orchestration, memory, tool calling, and end-to-end traces.
Start on the free Starter tier to prototype, move to Pro when you ship multi-agent systems to production, and graduate to Enterprise for SSO, SOC 2 audit logs, and dedicated infrastructure. Whatever the tier, you get the same SDKs, the same integrations, and the same platform features — the limits and guarantees are what scale.
Simple pricing that scales with you
Start free. Upgrade when your agents are doing real work.
Starter
For builders exploring agents.
- 1 active agent
- 500 runs / month
- 20 integrations
- Community support
- Basic observability
Pro
For teams shipping in production.
- Unlimited agents
- 25,000 runs / month
- 200+ integrations
- Multi-agent teams
- Full traces & replay
- Priority support
Enterprise
For scale, security & control.
- Everything in Pro
- Unlimited runs
- SSO / SAML & SCIM
- SOC 2 & audit logs
- Dedicated infrastructure
- 24/7 support + SLA
Compare plans feature by feature
The full breakdown of what each AI agent platform tier includes — from run limits to security and support.
| Feature | Starter | Pro | Enterprise |
|---|---|---|---|
| Active agents | 3 | 25 | Unlimited |
| Monthly runs | 1,000 | 100,000 | Custom |
| Integrations | Core | All 200+ | All + custom |
| Multi-agent teams | |||
| Traces & replay | 7-day | 90-day | Custom retention |
| SSO / SAML & SCIM | |||
| SOC 2 & audit logs | |||
| Support | Community | Priority email & chat | Dedicated TAM |
| Uptime SLA | 99.5% | 99.9% | |
| Dedicated infra / VPC |
What counts as a run?
One run is a single complete execution of an agent — the full perceive → reason → act → observe loop, including every tool call and reasoning step until the goal is met or a stop condition fires. Background sub-agents spawned by an orchestrator in the same task share that parent run, so a multi-agent workflow is not penalized for delegating.
What every tier includes
No paywalls on the fundamentals. The core agentic platform is the same whether you're on the free plan or Enterprise.
Full orchestration
Build single agents or coordinated teams with the same loop, retries, and guardrails — no feature flags hiding the good parts.
Memory & vector store
Short-term context plus long-term memory backed by a managed vector store so agents recall facts, past runs, and user preferences.
Tools & integrations
Connect databases, APIs, and SaaS apps. Starter gets core tools; paid tiers unlock the full integration catalog and custom tools.
Traces & replay
Every token, decision, and tool call is logged so you can debug, replay, and trust exactly what your agent did in production.
TypeScript & Python SDKs
Prototype visually, then drop into code. The same SDKs power every plan, with no artificial rate caps on the free tier.
Security baseline
Encryption in transit and at rest, scoped API keys, and per-tool permissions on every plan — with SOC 2 and SSO on Enterprise.
Pricing that scales with your results
Run-based billing means a more capable agent doesn't automatically cost more — efficiency is rewarded, not punished.
Free runs / month
on the Starter plan
Integrations
tools & data sources
Annual savings
vs. monthly billing
Enterprise SLA
uptime guarantee
Because you pay per run rather than per seat, an entire team can collaborate on the same workspace without inflating your bill, and you can give every engineer access to the full platform. When you need higher volume, audit logs, or a security review, our sales team will tailor an Enterprise plan to your run volume and compliance requirements.
AI agent platform pricing, answered
Yes. The Starter plan is free forever and is designed for students, side projects, and proof-of-concept work. It includes up to 3 active agents, 1,000 agent runs per month, core integrations, and full traces so you can debug every step of the agent loop. No credit card is required to start, and you can upgrade the moment you outgrow the limits.
Start building agents free
Spin up your first autonomous agent in minutes on the free plan — upgrade only when you ship to production. No credit card required.