Explain your task in plain English.
The agent interviews you about the task and writes a SPEC.md that drives every downstream stage.
Private beta
Finetune a custom LLM in minutes, with just a plain-English description of your task.
No need to bring data, evals, or rubrics. Lqh will help you create the right data and evaluations to fine-tune a custom model tailored to your needs.
$ curl -LsSf https://astral.sh/uv/install.sh | sh # install uv (one-time) $ uv tool install lqh $ lqh
How it works
Walkthrough the fine-tuning pipeline.
Lqh creates data for you, or bring your own
Lqh authors a per-task data pipeline, generates samples concurrently, and scores each one with an LLM judge against your rubric. The dataset that hits training is already curated.
Hands-off with --auto
Point lqh at a directory and walk away. It either delivers a checkpoint that beats baseline or returns an explicit failure with the reason.
The value of lqh
The official tool for fine-tuning Liquid Foundation Models.
Why use a custom fine-tuned LLM?
Fine-tuned LLMs are more efficient and cost-effective for high-volume tasks than commercial APIs, and perform with more specialized knowledge, better accuracy, and strict privacy. Custom training ensures output matches exact tone, formats, and brand voice.
Why customize a Liquid Foundation Model?
Our models are small enough to run on device, but aren't useful for every generic use case out of the box. Customization gives you the intelligence you need for the tasks that matter to you at a price point that can actually scale.
What sets lqh apart?
Other fine-tuning platforms assume you have perfect data and evals, but we know that is the hardest part.
Our platform addresses the real bottleneck: starting from a simple English description of your task, we help you create the right data and evaluations to fine-tune a custom model tailored to your needs.
Watch our demo
From description to model in one run.
Request access
Customize your first model in minutes.
LQH is in private beta. Drop your email and we'll send an install link.