Pay one flat monthly price, and that's the entire bill from us. No per-request fees, no cut of what you save. Trim hundreds off your provider bill and the number we charge you never moves.
See your spend. Start cutting it.
Up to 75% off input on every long conversation.
By the calculator below, Pro pays for itself around $110/mo of chat-style LLM spend. Clear that and the savings are all yours, every month.
One flat price for every client bot you run.
Runs in your own cloud. Your keys, your data.
Your dashboard tracks every request live, so you see exactly what you're saving as it happens. No card to start - your $5 credit runs on your own traffic.
| Free | Pro | Agency | Enterprisecoming soon | |
|---|---|---|---|---|
| Billing | ||||
| Base price | $0 | $29 / mo flat | $199 / mo flat | Custom flat license |
| Savings kept by you | 100% | 100% | 100% | 100% |
| Usage fees | None | None | None | None |
| Trial allowance | $5 compression credit (one-time) | n/a | n/a | n/a |
| Annual price | no | $290 / yr (2 months free) | $1,990 / yr (2 months free) | Annual contract |
| Limits | ||||
| API keys | 1 | 5 | 20 | Unlimited |
| Rate limit | 60 req/min | 300 req/min | 1,500 req/min | Unlimited |
| Tokens proxied | Unlimited | Unlimited | Unlimited | Unlimited |
| Features | ||||
| Streaming, tool use, vision | included | included | included | included |
| Response cache (repeats served at $0) | no | included | included | included |
| Concise output mode (beta) | no | included | included | included |
| Zero-retention mode | no | included | included | default |
| Spend caps | dashboard only | per-key | per-client | included |
| Cancel anytime via portal | n/a | included | included | contract |
| Self-hosted Docker | no | no | no | included |
| Usage | ||||
| History retention | 7-day | 90-day | 90-day | Configurable |
| Compression on long conversations | $5 free credit | Included | Included | Included |
| Production use | included | included | included | included |
| Support | ||||
| Channel | Community | Priority email | Dedicated + SLA | |
Yes. The flat price is the only thing PromptCrunch ever charges. No per-request fees, no percentage of savings, no overage, no credit packs. You still pay your LLM provider directly for the (now smaller) bill.
Free accounts get a $5 credit: we compress your traffic until you've saved $5 off your provider bill, plus 100 optimized requests a day. No card required. You see real dollar savings on your own prompts before you decide.
No. Code, JSON, schemas, IDs, and numbers come through verbatim - only redundant prose gets trimmed. On a 40-prompt Claude Sonnet benchmark the input bill dropped from $2.56 to $0.64, a 75% cut, with identical responses.
Neither hurts. Requests already hitting your provider's cache pass straight through, so you're never billed a cent worse than going direct. Caching wins when you resend the same prefix; PromptCrunch wins when the conversation grows. They stack.
Yes. Cancel from the dashboard via the billing portal. You keep service through the paid period and drop to Free.
A custom flat annual license for the self-hosted build. Same principle: one flat fee, you keep 100% of savings. Details here.
Yes. We're a proxy, not a reseller. Your key goes to the provider, never stored on our side.
Zero-retention mode on Pro and Agency holds no state on our servers. For zero-trust, the self-hosted build runs in your VPC.
Free to start, no card. Every dollar you save stays yours.