cashcrown // Models
Loading measurements…
real-time interfaces, fast feedback, short answers
The OpenClaw router picks a model per task — the cheapest that can carry it. This model is:
Run this model on your own prompt — in a safe sandbox (PII masked, zero retention).
An NVIDIA family with very low TTFT (~0.4s) — when an instant response start matters. Best for: real-time interfaces, fast feedback, short answers.
Yes — we serve it in production through the OpenClaw router, with PII masked before the cloud and measured parameters (see the live stat band).
The OpenClaw router picks this model for tasks: reasoning (fallback), fast answers (fallback), classification (fallback), extraction (fallback), summarization (fallback).