What happened
Giotto AI, a Lausanne-based research lab, launched two new AI models: Giotto Alto, a flagship reasoning model available for free via web and mobile apps with European hosting, and Giotto Solo, a single-GPU model for on-premise enterprise deployment. The launch was accompanied by immediate criticism regarding the models' origins, with users alleging Alto is based on China's Kimi model, and concerns over a license that prohibits independent benchmarking without permission.
On October 27, 2026, Giotto AI unveiled Giotto Alto and Giotto Solo. Alto is described as a flagship reasoning model running on eight Nvidia B300 GPUs, available for free in the browser and mobile apps with servers located in Europe. Solo is designed to run on a single GPU for organizations requiring on-premise AI solutions. The company, founded in 2017 by EPFL alumni, previously sold paid licenses to Swiss public bodies and companies.
The launch was met with immediate skepticism. Within hours, users on X alleged that Giotto Alto is actually a service running on Moonshot AI's Kimi model, a Chinese open-weight model. Critics pointed to instances where the model identified itself as Kimi when prompted. Additionally, researcher Alexander Doria highlighted license clauses that ban reverse engineering and prohibit publishing results without Giotto's written consent, effectively preventing independent verification of the company's performance claims.
Giotto AI has not disclosed the parameter count or for either model, describing the architecture as a coordinated network of specialized models. The company claims Alto ties the best scores on AIME 2026 but trails OpenAI's GPT-6 Astra and Claude Fable 5.1 on other difficult reasoning tests. No independent rankings have yet included Giotto's models. The free tier allows anonymous use, while business use and API access with higher parallelism are paid, though specific pricing figures were not provided in the launch materials.
The 'sovereign' positioning emphasizes European data residency and independence from US cloud providers like Microsoft, Amazon, and Google. However, the ambiguity surrounding the model's training data and weights contrasts with competitors like Mistral, which publishes open weights. This distinction is significant for entities complying with the European AI Act, where transparency about model provenance and capabilities is increasingly critical for regulatory and security assessments.
Source details: pasqualepillitteri.it
Why it matters
This launch represents a significant move in the European sovereign AI market, offering a free, accessible alternative to US-dominated cloud services while emphasizing data residency. However, the controversy over the models' lineage and the prohibition on independent testing raises critical questions about transparency and the true nature of 'sovereignty' in AI. For organizations, the on-premise option provides a path to avoid US cloud dependencies, but the lack of open weights and independent verification complicates compliance and trust assessments for regulated industries.
The launch offers a tangible alternative for European organizations seeking to reduce dependence on US-based AI infrastructure. The free access to Alto lowers the barrier to entry for testing European-hosted AI, while Solo provides a viable on-premise option for data-sensitive industries. This is particularly relevant for public sector entities and regulated companies that must ensure data sovereignty and compliance with GDPR and the AI Act.
However, the controversy over the models' origins challenges the core value proposition of 'sovereign AI.' If the models are derived from Chinese open-weight models, the claim of European sovereignty is undermined, especially regarding the potential influence of foreign training data and biases. The restrictive license that bans independent benchmarking further erodes trust, as it prevents the community from verifying performance claims and model integrity, which is a standard practice in the open-source AI ecosystem.
This incident highlights the tension between commercial secrecy and the transparency required for sovereign AI. While Giotto AI offers a practical solution for data residency, the lack of open weights and independent verification means that customers must rely on the vendor's word for both performance and provenance. This creates a risk for organizations that need to audit their AI supply chain for security and compliance reasons.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
Why can ethical evaluation not be reduced to one model score?
What to watch next
Monitor Giotto AI's response to the Kimi model allegations and any updates to their licensing terms. Watch for independent results from third-party evaluators, as current performance claims are vendor-reported. Observe adoption by Swiss public bodies and whether the 'sovereign' label holds up under regulatory scrutiny regarding model provenance and transparency.
Giotto AI's official response to the allegations that Alto is based on Kimi will be crucial. A clear explanation of the model's architecture and training data, or a commitment to open weights, could mitigate the controversy. Conversely, continued silence or vague statements may lead to a loss of credibility among technical users and regulators.
Independent benchmarking results are expected to be a key indicator of the models' true capabilities. If third-party evaluators can obtain access and publish results, it will provide a more reliable picture of how Alto and Solo compare to leading US models. The current vendor-reported scores are insufficient for making informed deployment decisions.
Adoption by high-profile Swiss public bodies, such as RUAG, will serve as a test case for the practical viability of Giotto's sovereign AI offering. If these entities continue to use the models despite the controversy, it may signal that data residency is prioritized over model provenance in certain contexts. However, any regulatory pushback or public criticism from these bodies could impact Giotto's reputation and future sales.