basal.
I use basal
BASAL BLOG

Models, results.
And what they mean in practice.

We write about basal releases, how the models work and applications worth trying.

Text goes into basal-2.0. Text, images and audio go into basal-omni. Both families return typed decisions through a shared API.
New release

basal-2.0 and basal-omni — more decisions per second, better calibration, images and audio

Three open basal-2.0 decision models for Polish and English: 1.4–5.2× more decisions per second than basal-1.5, better-calibrated probabilities, up to 255 options per question and ports with their own calibration. Alongside them, experimental basal-omni: two Gemma 4-based models that also read images and audio. We also explain where the new models do not improve, and how they rank in Werdykt.

17 min readRead article
New release

basal-1.5 — three models, more possibilities

A dynamic classifier defined in every request, without additional training. Three sizes, new response schemas, faster multi-question processing and full Werdykt benchmark results — including Polish.

10 min readRead article
Meet the model

basal-1.0 — a dynamic classifier

An open classifier for Polish and English. How it works, what it can do and how long a decision takes — from a B300 server to a desktop HP ZGX Nano (Nvidia DGX Spark).

7 min readRead article
FROM READING TO BUILDING

Have a process waiting for a decision?

Explore sample requests and responses for 18 classifier use cases.

Open the use case catalog