Pricing
One backend, four plans. Bring your own provider keys; each plan carries a monthly request and token allowance. Pick the tier that matches your traffic.
Free
$0
Evaluate and prototype.
- Requests / mo10K
- Projects1
- Log retention7 days
- Team seats1
- Tokens1M / mo
- Generative mediaNot included
- SupportCommunity
Hobby
$24.99/ mo
Ship a real app, solo.
- Requests / mo500K
- Projects5
- Log retention60 days
- Team seats1
- Tokens50M / mo
- Generative mediaImage · audio · video (your keys)
- SupportEmail
ProPopular
$99/ mo
Scale with traffic and a team.
- Requests / mo5M
- Projects25
- Log retention1 year
- Team seats5
- Tokens500M / mo
- Generative mediaImage · audio · video (your keys)
- SupportPriority
Enterprise
Custom
Discuss your scale and support requirements.
- Requests / moCustom
- ProjectsCustom
- Log retentionCustom
- Team seatsCustom
- TokensCustom
- Generative mediaImage · audio · video (your keys)
- SupportBy agreement
Cloud signup is open. Swift SDK repository access is a separate private beta; contact hello@agkit.cloud to request access.
Every plan is bring-your-own-keys: your provider bills you for tokens, never us — the monthly token allowance on each card is this backend's own ceiling, not a bill. Generative media routes to your own media keys under a per-project spend budget. It isn't a charge we collect.
Questions
- What counts as a request?
- A relay inference call is metered as a request. An agent task can involve several inference calls as it uses tools and continues. Application tool execution and provider inference are different operations; do not equate one user task with one request.
- Why bring your own keys?
- You connect your own Anthropic, OpenAI, Gemini, and media-provider keys. Your provider bills you directly for tokens, so we never mark them up. You pay only the provider's cost, and our price is for the backend.
- Is there a token limit?
- Yes. Each plan carries a monthly token allowance, shown on its card, and a call is refused once the month's usage reaches it. Your provider still bills you for the tokens themselves; the allowance is the backend's own ceiling. Per-request ceilings you set on your agent profiles protect you from a runaway loop inside a single call.
- What is log retention?
- How long your usage and audit history stays queryable in the dashboard before it's purged. Each plan card shows its own window.
- How does generative media work?
- Supported image, audio, and video requests route through your configured media providers under project controls. Consult the current plan cards for media availability. Provider charges are separate from Cloud fees.
- What happens above Pro?
- Contact us about your requirements and a written agreement. Custom service commitments are negotiated separately; draft legal pages are not an active service commitment.
- Can I change plans?
- Yes. Upgrade or downgrade any time; the new limits apply from the change.