Ships our productStac becomes the AI behind it

Claude and OpenAI charge per token, so your bill grows with the size of every prompt and every response. With Stac, you pay a fixed monthly price that covers a volume of requests, and each request counts once, whether it's short or long. You know what you'll spend before you start.

Go includes 4,000 requests per month and Pro includes 10,000. On Enterprise, the volume is set with you. If you run out before the month ends, you can turn on extra usage and pay only for the requests over your quota, always per request, never per token. Without extra usage, requests stop when the quota runs out and resume when it renews the following month.

Yes, among the models in your plan. Go includes Qwen 3.5 9B and GLM 5.3 Flash, and Pro adds Qwen 3.8 27B, our most capable model for code and agents. On Enterprise, the model is defined with you.

Pro has a larger quota (10,000 requests per month versus 4,000 on Go), adds Qwen 3.8 27B, accepts up to 256K tokens of context (128K on Go), and handles more simultaneous requests. Enterprise adds a dedicated GPU, fine-tuning on your own data, a DPA, full audit logs, and priority support with an SLA.

Yes. On annual billing, Go and Pro come with 7 free days. We ask for a card at sign-up and only charge at the end of the trial. Cancel before then if you don't want to continue.

Yes. If you want to fully isolate different products or use cases, you can subscribe to more than one Stac plan at the same time. Each subscription is independent, with its own keys and knowledge base. If what you need is a more capable model, the way to go is upgrading your plan.

No. You get an API key ready to integrate into your product, with no need to set up a server, GPU, or infrastructure.