FAQ
The short version lives on the waitlist. This is the full set - product, savings, security, integration, and early access.
Product
What is Prune?+
Prune is a managed AI gateway. Change one base URL so every call to OpenAI, Claude, Gemini, Bedrock, or OpenRouter goes through Prune - for unique-path cost control, Shield security, and a receipt on every response. Cache helps on repeats; it is not the product thesis.
How is Prune different from cache-focused AI gateways?+
Most AI gateways lead with cache hit rate. Prune leads with unique-path savings (routing, output caps, prompt cleanup) that apply when the prompt never repeats, plus Shield (vaulted keys, PII/secret checks, spend caps) and a machine-readable receipt. Cache is measured separately as upside.
Who is Prune for?+
Developers shipping vibe-coded AI apps who need a seatbelt: cache and cheaper-model routing, vaulted keys, Shield, and a signed receipt - without standing up their own proxy. If you want to operate LiteLLM yourself, do that. If you want one URL change and no proxy ops, use Prune.
How does Prune differ from LiteLLM?+
LiteLLM is an open-source, self-hosted proxy you operate yourself. Prune is a managed service: we run the infrastructure, encrypted key vault, spend caps, and caching. Use LiteLLM when you need full control on your own servers; use Prune when you want one URL change and no proxy ops.
Do I need to self-host Prune?+
No. Prune is a managed gateway. You point your SDK at our base URL. There is no proxy to deploy, scale, or patch. If you need to run the control plane on your own machines, LiteLLM is the better fit.
Savings
How does Prune reduce LLM costs?+
On every call Prune can route to a cheaper model when safe, enforce output limits, and trim wasted prompt text. When traffic repeats, exact and semantic cache add extra savings. Receipts break out unique-path vs cache savings so you are not sold a vanity hit rate.
Does Prune only help when prompts repeat?+
No. The non-negotiable thesis is that Prune must still win on unique traffic. Caching and semantic reuse are multipliers on top, shown separately in prune_metadata.
Will this slow down my API calls?+
Cache hits can be faster than upstream. A miss / unique-path call adds small overhead - published miss-path latency targets are in the docs. Spend caps run in under 10ms before the provider call.
Security
Is Prune secure?+
Provider API keys are stored in an encrypted vault (AES-256-GCM). Shield can block PII, secrets, and runaway spend before the provider call. Prune is in early access and is not SOC2-certified; evaluate architecture, not badge theater.
Where are provider API keys stored?+
Provider keys never live in your app or repo. They are stored in Prune’s AES-256-GCM encrypted vault, scoped per account and key. Your client only holds a Prune API key that authorizes calls through the proxy.
Does Prune store my prompts?+
Prune stores usage metadata (model, token counts, cache hits, savings estimates) for your dashboard. Full prompt text is not retained in analytics logs by default. Cached responses are scoped to your account and derived from hashed payloads - not shared across tenants.
Is Prune SOC 2 certified?+
Not yet. Private beta is not SOC 2. We publish the architecture (AES-256-GCM vault, per-account isolation, Shield before the provider call) so you can evaluate the system instead of a badge.
Integration
How does the proxy work?+
You point your existing OpenAI/Claude/Gemini SDK at Prune’s base URL (or use the Prune SDK). Prune vaults the provider key, applies spend caps and Shield checks, optionally optimizes the call, then forwards to the provider and returns the response with a signed receipt.
What is in a signed receipt?+
Each response can include prune_metadata: cost with and without Prune, unique-path vs cache savings, tokens saved, optimizations applied, and an Ed25519 receipt_signature with a signing key id - so finance and auditors can verify the numbers.
Which providers does Prune support?+
Production-proven: OpenAI, Anthropic (Claude), Google Gemini, OpenRouter, and AWS Bedrock via drop-in base URL or Python/TypeScript SDKs. Ollama works for local/dev through a tunnel. More providers are on the roadmap; do not assume equal readiness for every name on the internet.
Can I keep using the OpenAI or Anthropic SDK?+
Yes. Point the existing SDK at Prune’s base URL and pass your Prune API key, or swap the import for the Prune SDK. Request shapes stay the same. To leave, change the base URL back - there is no lock-in.
How do I verify a Prune receipt?+
Every response can include prune_metadata. When receipt signing is configured, paste the object at https://www.withprune.com/verify or use the SDK verify helpers against the published Ed25519 public key.
How long does it take to integrate?+
Typically under five minutes: create a Prune API key, vault your provider key, point your SDK at Prune, send one request, and read the receipt.
What happens if Prune goes down?+
If Prune is unreachable, your app cannot reach the provider through the proxy until service is restored. For production workloads, keep a fallback path (direct provider URL) behind a feature flag. Uptime targets will be published for paid plans; status updates go out during incidents.
Beta & pricing
What happens after I get early access?+
Create your account with the same email, vault a provider key, send one request, and read the signed receipt. Save (cache + routing), Shield, Prove. Reach hello@withprune.com if you get stuck.
How much does Prune cost?+
Early access runs on Free (15,000 proxied requests/mo or 14 days from your first Optimize receipt). Planned packaging when paid checkout opens: Starter $15/mo (250k), Pro $49/mo (1M), Team $149/mo (5M). You still pay your LLM provider for model usage. Prune reduces that bill via cache, routing, and unique-path savings.
Still comparing? vs cache-first gateways · how it works · help center
Ship the next AI feature with a seatbelt already on.
Get early access
