Apertus 1.5 adds 4T tokens to 8B model, 2T to 70B, with multimodal and long-context upgrades
Apertus released version 1.5 of its open-weight models, adding 4 trillion tokens to the 8B model and 2 trillion to the 70B model. The update expands context to 262,144 tokens, introduces native multimodal understanding (images and audio alongside text), and improves reasoning with a ‘thinking mode.’ The models also better follow instructions and integrate with external tools, while adhering to…
Key points
- Added 4T tokens to 8B model and 2T to 70B model, totaling 19T and 17T tokens respectively
- Context window expanded to 262,144 tokens (4x increase) with native multimodal support for images/audio
- Outperforms similarly sized open models in coding, multilingual tasks, and legal reasoning; API access via Swisscom, Infomaniak, and others
The release includes detailed training data, intermediate checkpoints, and benchmark results published under the Every Eval Ever (EEE) schema. Apertus 1.5 outperforms similarly sized open models in coding, multilingual tasks, and legal reasoning, with multimodal performance competitive with closed-source models like Gemma 4 12B. Swisscom, Infomaniak, and Phoeniqs have announced API access, and apps like Proton Lumo and Infomaniak Euria have integrated the models. The project also previewed Apertus 2.0, targeting a sparse/MoE architecture with ~600B parameters.
Apertus 1.5 Release Notes
Hugging Face · 15 October 2026
Loading the full article…
This text was published by Hugging Face. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Clash of AIs platform shows ChatGPT leads Claude in market tests · 1 src
- Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents · 30 src
- Google launches Gemini 4 Argon model globally for Pro and Ultra subscribers with introductory API pricing · 29 src
- SB Intuitions releases Sarashina free beta; translation test shows natural Japanese but meaning shifts · 1 src
- TypeSafe AI releases Jev 1.14 with images, box type and 262k token context · 1 src
Comments
via GitHub Discussions