Give Laya a state (text or JSON) and typed questions; it returns typed answers with calibrated probabilities in a single forward pass (~35 ms). No text generation — nothing to parse, nothing to hallucinate. Model: convaiinnovations/laya (421M, Apache 2.0).
static demoinference via official API This page runs entirely in your browser and calls the official laya-demo Space over its public API. If you hit quota limits, open that Space while logged in and retry.
Classify, detect urgency, score frustration and check for a refund request in one call, then route.
Results appear here.
Screen every prompt before it reaches your expensive model: jailbreak, injection, sensitive data, harm.
Results appear here.
One Router, three checkpoints. The script of the incoming text decides which one answers — before the forward pass, because the model's own confidence gives no warning when it cannot read the input.
Results appear here.
Any state, any questions — the same request shape as the API.
Results appear here.