With 1.5x the love, from Switzerland
Apertus 1.5 makes Switzerland's open language models more useful in practice: multimodal input, reasoning, longer context, and better instruction following.
July 24, 2026
On July 24, 2026, the Swiss AI Initiative—a collaboration between EPFL, ETH Zurich, and the Swiss National Supercomputing Centre (CSCS)—released Apertus 1.5, a major update to the country's fully open language models. Building on Apertus 1.0, the 8B and 70B models are now more capable in the ways that matter for real-world use, while staying true to the same values: transparency, multilingualism, and public service.
Apertus 1.5 is live today on the Public AI Inference Utility and available wherever you build.
- Chat with Apertus 1.5 (this is the 70B reasoning model)
- Developer platform
- Models on Hugging Face
- Apertus 1.5 announcement
- Chat with Apertus 1.5, no login necessary (this is the 8B reasoning model)
Public AI is proud to be the official international deployer for Apertus. Apertus 1.5 is available now on chat.publicai.co, platform.publicai.co, and Hugging Face, the same open infrastructure that brought Apertus to the world last September, now serving the next generation of the model.
What's new in 1.5
Apertus 1.5 is a continued pretraining of Apertus 1.0: the 8B model adds 4 trillion tokens of text and multimodal training data; the 70B model adds 2 trillion. As with every Apertus release, the models remain fully open—open weights, open data, open values, and full training details.
A technical report with benchmark results, training pipelines, and intermediate checkpoints will follow in the coming weeks. For now, the model cards on Hugging Face have everything you need to run the models yourself.
See and reason
Apertus 1.5 accepts images alongside text, bringing documents, diagrams, and photos into the conversation. The models can also process spoken language, though this remains experimental.
An optional thinking mode lets the model reason about the input before answering—improving performance on problems that benefit from a step back before a reply.
Built for real use
The context window grows to 262,144 tokens, a four-fold increase over Apertus 1.0. Instruction-following is improved, giving more predictable and accurate responses. Tool use is better too: the models are trained for more effective integration with external tools and APIs.
Together, these updates make Apertus more useful for the applications people actually build, including summarization, tutoring, document analysis, agent workflows, and more.
Open by design
The name "Apertus" comes from Latin, meaning open. Apertus 1.5 strengthens adherence to the Apertus Charter (formerly known as the Swiss AI Charter), which brings transparency to the model's values and principles. Everything remains inspectable and reproducible:
- the training architecture, datasets, and recipes
- the model weights, including intermediate checkpoints
- the source code, logs, and deployment guides
Unlike models that offer partial access, Apertus is fully inspectable and modifiable, giving developers, researchers, and institutions complete visibility into how the model was built and how it behaves.
Try Apertus 1.5 now
Wherever you are, you can try Apertus 1.5 right here:
Where it goes next
The Swiss National AI Initiative will continue to make regular updates to Apertus, exploring domain-specific models in areas like law, climate, health, and education. Apertus 1.5 is the next step. Expect Apertus 2 to come out in months, not years.
The Public AI Inference Utility will continue to bring Apertus to the world. We look forward to supporting more Apertus-scale efforts, and we extend an open invitation to other public institutions, research communities, and forward-looking companies to join us in shaping the next chapter of AI.