Switzerland Launches Open-Source AI Model Apertus
Key Points
- Switzerland released an open‑source AI model named Apertus.
- Apertus is positioned as an alternative to ChatGPT and Claude.
- The model’s code, data, weights and development details are on HuggingFace.
- Training covered over 1,800 languages.
- Two sizes are offered: 8 billion and 70 billion parameters.
- Performance is said to be comparable to Meta’s Llama 3 model.
- Training data was limited to public sources that allowed crawling.
- The project complies with EU copyright law and respects opt‑out requests.
Switzerland has introduced an open‑source artificial intelligence model named Apertus, positioning it as an alternative to proprietary systems such as OpenAI’s ChatGPT and Anthropic’s Claude. The model’s code, training data, weights and development details are publicly available on HuggingFace. Designed to establish a new baseline for trustworthy, globally relevant AI, Apertus was trained on more than 1,800 languages and is offered in two configurations—one with eight billion parameters and another with seventy billion. The developers say the model aligns with European Union copyright rules and respects opt‑out requests from websites, avoiding any stealth‑crawling of data.
Open‑Source Ambition
Switzerland unveiled an open‑source artificial‑intelligence model called Apertus, promoting it as a viable alternative to well‑known proprietary offerings like OpenAI’s ChatGPT and Anthropic’s Claude. By publishing the model’s source code, training datasets, model weights and a detailed development roadmap on the HuggingFace platform, the Swiss team emphasizes transparency and community collaboration.
Design Goals and Scope
The developers describe Apertus—Latin for “open”—as an effort to set a new baseline for trustworthy and globally relevant open models. To achieve this, the model was trained on a multilingual corpus spanning more than 1,800 languages, reflecting a broad linguistic ambition that exceeds many existing commercial systems.
Model Sizes and Capabilities
Apertus is released in two configurations. The smaller version contains eight billion parameters, while the larger version scales up to seventy billion parameters. According to the developers, the larger model’s capabilities are comparable to Meta’s 2024 Llama 3 model, indicating a competitive performance profile within the open‑source arena.
Compliance with European Regulations
The project was built to comply with European Union copyright legislation and the voluntary AI code of practice that several U.S.‑based AI firms have signed. Training data was sourced exclusively from public websites that allowed crawling, and the team respected opt‑out requests from sites that declined participation, thereby avoiding any “stealth‑crawling” practices.
Strategic Context
By offering a fully open model, Switzerland aims to provide an alternative path for organizations and developers who seek AI capabilities without reliance on proprietary platforms. The availability of the model on HuggingFace also facilitates community‑driven improvements, potential customizations and broader academic research.
Reception and Outlook
Industry observers note that Apertus joins a growing list of open‑source AI initiatives that challenge the dominance of commercial giants. Its multilingual breadth and adherence to European legal standards may attract users focused on data sovereignty and ethical AI practices. The model’s launch marks a notable step for‑ward for European contributions to the global AI ecosystem.