rebyte.ai

Rebyte Model Proxy

Rebyte Model Proxy

One API key per organization for direct model calls. Use an OpenAI-compatible endpoint, route through Rebyte, and charge usage to the same organization credits that power agent runs.

Create a model key See the request
01

OpenAI-compatible by default

Drop one organization key into any backend.

Rebyte model keys are raw proxy keys for server-side use. The call goes directly to the model proxy; Rebyte attributes usage from the key metadata and deducts credits from the organization balance.

curl 
curl https://model.rebyte.ai/v1/chat/completions \
  -H "Authorization: Bearer sk_your_model_proxy_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4-mini",
    "messages": [{"role": "user", "content": "Hello"}]
  }'
02

Built for organization billing

01

Organization-owned keys

Create and revoke keys from the API Keys settings page. Keys belong to the current organization, not an individual user session.

02

Credit-gated requests

The model proxy checks available credits before provider calls and records usage back to the organization ledger.

03

Provider routing behind one endpoint

Use one endpoint while Rebyte handles model naming, provider credentials, and routing behind the proxy.

03

Request path

1

Your server calls model.rebyte.ai/v1 with the model key.

2

LiteLLM reads the organization metadata attached to the key.

3

Credits are checked, provider usage is logged, and the organization balance is updated.

Give teams direct model access without handing out provider keys.

Start with one organization-scoped key, then rotate or revoke it from Rebyte when access should stop.

Open API Keys
rebyte.ai · san francisco · 2026
Registrati →