
Routehook
The AI gateway for builders: one key and one bill for 50+ AI models.
One API, one key and one prepaid balance for text, image, video, audio and embedding models from every major provider, priced below what the providers charge.
- Status
- Live
- Category
- AI API gateway
- Years
- 2026
In numbers
50+
models behind one key
11
provider adapters behind one endpoint
3
wire formats: Chat Completions, Responses, Anthropic Messages
5
routing strategies per model
The story
The problem
Using several AI providers means several integrations, keys and invoices, and switching models usually means rewriting code. Image and video models add another problem: a generation can run for minutes, fail halfway, and still has to be billed correctly.
What was built
A gateway that speaks the OpenAI and Anthropic wire formats, with provider adapters behind it, async jobs for image and video, per-key rate limits, routing and failover between providers, a hold-then-settle billing ledger, a customer dashboard with a live playground, and an admin console.
Highlights
What it does best.
Drop-in compatible
Keep the OpenAI library you use. Change the address and the key, and that’s it.
Below list price
Every model priced next to the provider’s own rate, so the savings are checkable.
Every kind of model
Text, image, video, audio and embeddings from OpenAI, Anthropic, Google, xAI and more.
Long jobs, handled
Image and video run as queued jobs with webhooks and failover.
Spend caps per key
A runaway bug stops at your limit, not on your bill. Never billed twice.
Private by default
Provider keys sealed with AES-256-GCM. Prompts are never used for training.
Who it’s for
- Developers building AI products
- Teams using several AI providers
System & tech
- Next.js
- NestJS
- PostgreSQL
- Redis
- Drizzle
- Stripe
- better-auth
- Docker
Where it stands.
Live at routehook.ai with a public model catalogue and published per-request pricing. No monthly fee: add $5 of credit and pay per request.
Sources: routehook.ai · Vibemonk case study
Connected

