Small LLMs on-device: what a 0.5B model can do
A small local language model is a good fit for narrow, text-in text-out jobs where the answer is already in the input. It is a poor source of knowledge, and it needs a person checking its output.
Speech is transcribed first, then a small instruct model drafts a structured clinical summary from the transcript — the same two-stage pattern behind our ambient scribe work, running entirely on-device here. Both models are small for browser-friendly size, so treat the formatting as a rough draft, not a polished SOAP note — production engagements use larger, evaluated models for that.
A small local language model is a good fit for narrow, text-in text-out jobs where the answer is already in the input. It is a poor source of knowledge, and it needs a person checking its output.
Send us the paragraph you'd normally write to a colleague. We'll get back to you within 24 hours.
hello@tathastha.com