Introducing Quasar 1.0 – our strongest model

Model family

Pico

How small can a language model be? The Pico family goes from 11,000 to 6.9 million parameters. Pico 1 is about 0.5 MB and runs even on a Raspberry Pi.

In numbers

Pico models
4
parameters in the smallest (Pico 1.3 Small)
11K
parameters in the largest (Pico 1.3 Coding Large)
6.9M
best score (QA Bonus)
2.2%

An experiment in size

Pico 1 has only 254,000 parameters. It can hold simple small talk in English, knows that it is a tiny AI and honestly says that it cannot do maths or write code. Its answers often don't fit the question.

Pico 1.2, Pico 1.3 Small and Pico 1.3 Coding Large explore the same question from both ends: even smaller, and a version focused on code.

All Pico models on Hugging Face:

Benchmarks

Convergent 3.0 scores, the same numbers as on the benchmarks page. Best of 10 runs per benchmark.

Pico 1.3 Coding LargePico 1.3 SmallPico 1.3 MathPico 1.2Pico 1.1
Agentic codingConvergent 3.0 · Basis0.0%0.0%—pending0.0%0.0%
Knowledge workConvergent 3.0 · Basis1.3%0.0%—pending0.25%0.05%
Question answeringConvergent 3.0 · Basis2.0%2.2%—pending0.0%0.0%
Agentic codingConvergent 3.0 · Bonus0.0%0.0%—pending0.0%0.0%
Knowledge workConvergent 3.0 · Bonus0.8%0.0%—pending0.7%0.1%
Question answeringConvergent 3.0 · Bonus2.2%0.0%—pending2.0%0.0%
Humanity's Current ExamConvergent 3.00.57%0.0%—pending0.2%0.04%
HTML codingConvergent 3.00.0%0.0%—pending0.0%0.0%
MathConvergent 3.0 · Basis2.2%0.0%3.0%4.0%3.7%
MathConvergent 3.0 · Bonus7.5%3.3%4.0%7.9%0.83%
Validation lossConvergent 3.0—not measured yet—not measured yet—pending—not measured yet—not measured yet
Parameters6.9M11K8.1M461K254K
Training hardwareASUS TUF RTX 3060 TiASUS TUF RTX 3060 TiASUS TUF RTX 3060 TiASUS TUF RTX 3060 TiASUS TUF RTX 3060 Ti

Score per category

Best of 10 runs per benchmark.

00.631.31.92.5CodingBasisCodingBonusKnowledgeBasisKnowledgeBonusQABasisQABonusHCEHTML codingScore (%)10K100K1M10MParameterslog scale11K461K254K001.30.8022.20.5706.9M

Try Pico 1

Pico 1 is public on Hugging Face.