VaultAPI

BYOK gateway · 6 providers · one endpoint

Your AI bill has
a control room now.

VaultAPI sits between your app and every LLM you use. Route through one endpoint, count tokens before you spend them, and let routing rules pick the cheapest model that can do the job.

# one endpoint, any model

POST /v1/chat/completions

{ "model": "gpt-4.1", "messages": [...] }

routing short prompt → gpt-4o-mini

tokens 41 in · 128 out

cost $0.000083 (−93% vs requested)

status 200 · logged · under budget

OpenAIAnthropicGeminiGroqMistralOpenRouter

Live in minutes, not sprints.

No SDK to learn, no proxy to deploy. Three steps and every AI request you make flows through your own control room.

  1. 1

    Connect a provider key

    Paste your OpenAI, Anthropic or Gemini key once. It's envelope-encrypted the moment it lands and only decrypted for the instant a request needs it.

  2. 2

    Mint a gateway key

    Create a vk_ key scoped to your workspace. Revoke it any time — it dies everywhere, instantly.

  3. 3

    Change one line

    Point base_url at VaultAPI and keep the SDK you already use. Every request is now routed, counted, costed and logged.

your_app.py — the whole migration

client = OpenAI(

- base_url="https://api.openai.com/v1",

+ base_url="https://<your-vaultapi>/v1",

- api_key=os.environ["OPENAI_API_KEY"],

+ api_key="vk_live_…", # your gateway key

)

# streaming, fallbacks & every provider — unchanged code

📤Your appOpenAI-style call
🛡️VaultAPIAuth · limits · budgets
🔀RoutingBest model for the prompt
🧠ProviderYour key, decrypted just-in-time
ResponseLogged · counted · costed

Everything between your code and the model, handled.

01

One gateway, six providers

OpenAI, Anthropic, Gemini, Groq, Mistral and OpenRouter behind a single OpenAI-compatible endpoint. Streaming, retries and cross-provider fallbacks handled for you.

02

Keys stay yours

Bring your own provider keys. We envelope-encrypt them with AES-GCM, and the plaintext never touches a log, a response, or a disk.

03

Count before you spend

Provider-exact token counts and a cost estimate for any prompt, across every model — before a single request leaves your app.

04

Smart model switching

Routing rules send light prompts to cheaper models automatically, and the advisor mines your traffic to suggest new rules with projected savings.

05

See every request

Tokens, latency, cost, retries and fallbacks — logged per request, charted per day, with budgets that alert at 50/80/100% or hard-stop traffic.

06

Made for teams

Workspaces, roles, audit trails and scoped API keys. Solo today, twenty engineers tomorrow — same gateway.

Three days free. No card up front.

Add a provider key, send one request, and watch the first log line appear.

Create your workspace