Apertus 1.5: From Open Text Models to Open Multimodal Reasoners
Apertus 1.5 adds image and audio reasoning to fully open 8B and 70B models with a 262K context.
Apertus 1.5 extends the fully open Apertus family from text to multimodal reasoning. Continued pretraining of the 1.0 models, 4 trillion additional tokens for the 8B and 2 trillion for the 70B, adds image and audio understanding and extends context fourfold to 262,144 tokens, without changing the architecture. A chat-template thinking flag raises mathematical reasoning by up to 20 points on the 8B. Weights, data-preparation and training recipes, and evaluation suites are released under permissive licenses, trained only on openly available data that respects crawler opt-outs.
- 8B trained on 4T more tokens; 70B on 2T.
- Context window quadruples to 262,144 tokens.
- Thinking mode lifts 8B math scores by up to 20 points.
- Weights, recipes, and evals released under permissive licenses.
From the original article
1 min read at huggingface.co
Language models are becoming infrastructure, and infrastructure built in the open, by public institutions, on documented data, is the alternative to dependence on a few providers; Apertus shows this is viable at scale. Apertus 1. 5 extends the fully open Apertus family from text to multimodal reasoning.
By Project Apertus, Yixuan Xu, Yassine Turki · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.