Show HN: I built a hypervisor and client for inference on consumer compute
Scalattice speaks the OpenAI Chat Completions shape. If your app already uses the official SDK, you do not rewrite the stack — you point it at our API and pick a catalog model ID.
1. Get a key
Open
Scalattice Cloud and sign in.
Create an API key from the developers dashboard.
Keep credits topped up (prepaid). Published rates are on
pricing.
2. Python (OpenAI SDK)
from openai import OpenAI
client = OpenAI(
base_url="https://api.scalattice.cloud/v1",
api_key="slt_…", # from Scalattice Cloud
)
resp = client.chat.completions.create(
model="qwen-3-8b",
messages=[],
)
print(resp.choices[0].message.content)
Other catalog IDs work the same way (qwen-3-14b, llama-3.3-70b, deepseek-r1-7b, and the rest on pricing / Cloud).
3. curl
curl https://api.scalattice.cloud/v1/chat/completions \
-H "Authorization: Bearer slt_…" \
-H "Content-Type: application/json" \
-d ']
}'
What you get beyond the drop-in
Published per-token rates — no opaque quote for standard catalog traffic.
Output vetting and security tiers — optional controls on the request path.
Regional policy — constrain where work can run when compliance needs it.
Full product context: developers. Hosting a GPU instead? Host inference on your GPU.
Checklist if something fails
Base URL must be https://api.scalattice.cloud/v1 (include /v1).
Model ID must match the live catalog (not an OpenAI model name).
401 → wrong or revoked key. 402 / insufficient credits → top up in Cloud.
Slow or empty capacity → try a smaller model (for example qwen-3-8b) or retry; the network is demand-routed.
Ready? Create a key and run the snippet. Compare live rates anytime.
