Drop-in shape
chat/completions, streaming, the client libraries you already use. Point base URL and key. No new SDK.
Inference API · flat rate
OpenAI-shaped chat completions. One monthly price. Capacity we operate. Credentials only after you subscribe.
ScrollProduct
Three rules. Same request shape. We own the hardware path so you do not rewire when capacity grows.
chat/completions, streaming, the client libraries you already use. Point base URL and key. No new SDK.
$297/mo for text. You are not metered per token. Fair-use caps keep shared capacity stable for every seat.
Public model id and key stay the same when we add GPUs. Your integration does not change with our rack.
Access
Account + Stripe checkout. No key before payment clears.
Base URL, model, limits, and key in one place. Member guide included.
Use model legia-max. Keep the key on the server. Never in a public frontend.
Request shape
Placeholder only. Your real base URL and key appear in the member dashboard after subscribe.
curl YOUR_BASE_URL/v1/chat/completions \
-H "Authorization: Bearer YOUR_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "legia-max",
"messages": [
{"role": "user", "content": "Hello"}
]
}'Add-on
Spoken replies for +$100/mo. Toggle in the dashboard. Same account, separate speech path.
Pricing
Limited seats on shared capacity. Built for product traffic, not resale scrapers.
Pro
Flat rate for chat completions. Personal key, usage counters, member guide.
Stripe checkout. Cancel in the customer portal anytime.
legia-max · streamingFAQ
Only in your dashboard after subscribe. Marketing pages never show production endpoints.
Yes. Same chat completions shape. Full examples live in the member guide, not here.
No invoice by token count. Fair-use rate limits protect capacity for every subscriber.
Keep the key on your server. Cap max tokens. Handle 401, 429, and 5xx with backoff.