Paradigm Releases Violetto, Its First 1B Math Model

Paradigm has released the technical report for its first model, Limite 1B - Violetto, and open-sourced it under Apache 2.0. It's a 1-billion-parameter small model trained from scratch on carefully curated math data, focused entirely on mathematical tasks. The goal is to push "math capability per parameter" to the limit—achieving the strongest possible mathematical reasoning with minimal parameters and energy consumption. This is the third model released in Europe within two weeks.

Confirmed

• The model is named Limite 1B - Violetto, Paradigm's first model, accompanied by a full technical report
• 1 billion parameters, trained from scratch, focused on mathematical reasoning, emphasizing math capability per unit of energy and per parameter
• Open-sourced under Apache 2.0
• The model's naming and slogan use an anime style; the slogan translates to "There is only one solution, and your win rate converges to zero"

Unconfirmed

• Reposts claim the model scored 94 on AIME 2026; this is secondhand information—refer to the technical report for actual results

Why it matters

• Mathematical reasoning in small-parameter models keeps being pushed further. If the results hold, it means 1B-scale models can already approach large models' competition-math performance, with direct implications for inference-side deployment costs
• Paradigm, with a crypto-industry background, crossing over into model releases—plus Europe's recent string of new models—signals rising momentum in the small-model/vertical-model space

2026-10-07 ~ 2026-10-07 · 5 related posts

• 1B model scores 94% on AIME 2026: Paradigma open-sources Limite 1B Violetto — tensorqt · 2026-10-07
• Paradigm releases tech report for Limite 1B-Violetto, a from-scratch math model — tensorqt · 2026-10-07
• Limite 1B Violetto tech report: trained from scratch to max out math per parameter — tensorqt · 2026-10-07
• Paradigm releases Limite 1B -Violetto technical report: a tiny model trained from scratch to max out math per parameter — tensorqt · 2026-10-07
• Paradigm releases tech report for Violetto, a 1B model trained from scratch for math — tensorqt · 2026-10-07