Hobby
For side projects and experimentation.
- 5K requests / month
- Community support (Discord)
- 1 project
- Shared inference pool
- Basic observability
Start free. Upgrade when your eval suite — not your vendor — tells you to. No hidden fees, no "contact us for pricing" on the core tiers.
For side projects and experimentation.
For small teams shipping their first AI feature.
For companies running AI in production at scale.
For regulated industries and massive scale.
Everything you need to know.
| Feature | Hobby | Team | Scale | Enterprise |
|---|---|---|---|---|
| Requests / month | 5,000 | 100,000 | 1,000,000 | Unlimited |
| Deterministic Replay | — | ✓ | ✓ | ✓ |
| Eval Suite | Basic | Full | Full + Synthetic Data | Custom |
| SSO / SAML | — | — | ✓ | ✓ |
| Uptime SLA | None | 99.9% | 99.99% | 99.999% |
A request is a single call to the `stack.run()` method. Internal tool calls, retrieval steps, and model routing within that single request do not count as extra requests.
Yes. You can upgrade or downgrade at any time from your dashboard. If you upgrade, the new rate applies immediately. If you downgrade, it applies at the start of your next billing cycle.
Yes. If you are a pre-Series A startup, we offer 50% off the Team or Scale tier for your first 12 months. Contact support@cognitivestacktech.com with your company details.
We won't drop your traffic. Overage is billed at $0.002 per 1,000 requests on the Team tier, and $0.001 on the Scale tier. You can also set hard limits in the dashboard.
Yes, both Team and Scale come with a 14-day free trial. No credit card required to start.