Apertus 1.5 Release Notes

Hugging Face has released Apertus 1.5, extending the 8B and 70B Apertus 1.0 models with image and audio understanding, an optional thinking mode, a four times longer context window, and improved tool use and instruction following. Released on July 24, 2026, under an Apache 2.0 license, the updated model weights build upon continued pretraining adding multimodal and text data. Multimodality is introduced via an early-fusion approach where images and audio enter as discrete tokens without requiring a separate vision encoder. Over half of the training tokens remain text-only to preserve original language capabilities, supported by refreshed datasets that adhere to permissively licensed and publicly available criteria. Post-training spans five stages, and evaluation scope has significantly expanded to include multimodal assessments, translation, safety, and thinking-mode benchmarks. Apertus 1.5 results are published using the Every Eval Ever schema to support transparent and reproducible benchmarking. Applications such as Proton Lumo and Infomaniak Euria have integrated Apertus 1.5, and API access is provided by companies including Swisscom and Infomaniak. Additionally, the team announced the roadmap for Apertus 2.0 at the AI+X Summit in Zurich, featuring a sparse/MoE architecture with approximately 600B total parameters.