Apertus v1.5 70B arrives with an Apache-2.0 license

The Swiss AI initiative has published Apertus v1.5 70B, a 70-billion-parameter model released on Hugging Face under the permissive Apache-2.0 license. It is described as a multilingual system with multimodal capability, handling both text and images.

The headline detail here is less about raw size than about openness. At 70B parameters, Apertus sits in the class of models large enough to be broadly capable but still runnable outside hyperscale data centers. Pairing that with an Apache-2.0 license means researchers and companies can fine-tune, redistribute, and build commercial products on top of it without the usage restrictions that accompany many so-called open releases.

Why it matters

Europe has been searching for credible, fully open alternatives to models from large US labs, and a Swiss-backed effort adds welcome geographic and institutional diversity to that landscape. Key points from the release:

• A dense 70B model (not a mixture-of-experts design)
• Multilingual coverage plus image understanding
• Truly permissive Apache-2.0 terms for commercial and research use

Some specifics remain unstated in the model card, including the context window and detailed benchmark results, so teams evaluating Apertus for production will want to test it against their own workloads. For an ecosystem that increasingly values reproducibility and clear licensing, though, a capable 70B model with no strings attached is a meaningful addition.