One key per project, isolated billing
Issue an Anvat key for each engagement. Per-key spend cap, per-key RPM cap, per-key usage report. Bill the client directly or mark up the included cost — your call.
Anvat for agencies
One gateway, every client. Bill them, keep the margin.
Anvat is the AI API back-end for agencies running multiple AI projects in parallel. Per-key isolation, per-project spend caps, white-labellable dashboards. 30% off list + 2× credit pack = real margin on resale or pass-through.
- OpenAI-compatible
- Anthropic-compatible
- MCP-ready
- One API key
- 30% off list
- 2× credit on prepaid
- No card to start
- 5 languages
Why agencies use Anvat
30% off list + 2× credit pack = real margin
Pay Anvat $1, deliver $1 of inference. Pass through to the client at provider list price and you keep the spread. Most agencies clear 50%+ on AI delivery without changing the line items they bill.
Every model under one integration
Claude Opus, Sonnet, Haiku, GPT-5.5, Gemini, DeepSeek — all one base URL, one auth scheme. Client wants 'cheapest workable model'? Swap with one config change. No 'we need to onboard a new provider' delays.
Per-request cost telemetry built in
Every response carries x-anvat-cost-usd. Pipe it into Linear, Notion, your invoicing system, or a spreadsheet — instant cost attribution per ticket or per client.
Where the math actually changes
| What hurts today | What changes on Anvat |
|---|---|
| Maintaining 6 different provider keys across 6 clients | One Anvat account, one key per client. Per-key isolation, per-key caps. |
| Margin destroyed because you priced AI work in 2024 dollars | 30% off + 2× credit pack rebuilds the margin. Pass-through or resell — your call. |
| One client's runaway batch job blowing up the shared OpenAI quota | Per-key concurrent + RPM caps. One client can't starve another. |
| Procurement asking 'show me the cost per ticket' — no clean answer | x-anvat-cost-usd response header. Pipe straight into invoicing. |
| Trying to white-label the dashboard for client reporting | Auditable per-key usage CSV export. Forward to client or wrap in your own dashboard. |
Recommended plan
Pro X5 or Ultra (depending on monthly volume)
Pro X5 ($750 monthly credit, higher RPM) suits agencies running 3-10 active AI engagements. Ultra ($3,000 credit) suits 10+ concurrent projects or single high-volume clients. Both let you issue isolated per-client keys under the same plan — no per-seat licensing.
See plan details →Questions agencies ask before signing up
→Can I issue separate keys for separate clients?
Yes. Create any number of API keys from /app/keys. Each one has its own spend cap, rate limit, and usage stream. Rotate or revoke instantly without affecting the rest.
→Do you support a white-label dashboard for client reporting?
Not natively yet — we expose per-key usage via the dashboard + CSV export today. A full white-label admin is on the roadmap; for now, most agencies pull the CSV monthly and wrap it in their own client portal. Email hello@anvat.app if you want this prioritised.
→Do you have a referral / partner program for agencies?
Yes. Every Anvat user has a referral code that grants both sides credit when used. We can extend bespoke rev-share for high-volume agencies — email us.
→We sometimes need to comply with client data residency requirements.
We do not store prompt or completion bodies on Anvat — only billing metadata (token counts, latency, cost). Upstream calls hit the provider's chosen region (Anthropic US, OpenAI EU, etc.) — that contract is still between you and the upstream. Talk to us about specifics.
→Can we resell Anvat as part of our managed-AI offering?
Yes. Pass-through reselling is fine — pay Anvat list-minus-30, charge the client whatever you bill them. Brand-resale (re-skinning the dashboard) requires a partner agreement; email hello@anvat.app.
Ready to cut your AI bill?
Free signup, $2 credit, no card required. The SDK doesn't change — only the bill does.