Integrate in Three Steps
Create an API Key
Set project-specific credentials, limits, and permissions.
Choose a Model
Match the model to your task, budget, and latency target.
Send a Request
Use one request format across applications and automated workflows.
Authentication
Pass Authorization: Bearer YOUR_API_KEY in the request header. Never expose credentials in browser code, public repositories, or logs.
Send a Request
The example below shows the integration pattern. Your production endpoint and model catalog are confirmed during onboarding.
curl https://api.tokenforge.ai/v1/chat/completions \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"gpt-4o","messages":[
{"role":"user","content":"Hello TokenForge"}
]}'Compatible Request Format
The API follows a familiar Chat Completions structure to simplify migration from existing applications.
Model Routing
Route by capability, latency, cost, and current availability. Model names indicate compatibility and do not imply official partnerships.
Error Handling
Use exponential backoff and structured logging for timeouts, quota errors, model unavailability, and rate limits. Enterprise plans can add allowlists and failover policies.
Integration Guide
Prepare estimated volume, peak concurrency, target models, deployment regions, and billing preferences. A technical advisor will then configure your account, keys, and routing.
Talk to a Technical Advisor