A Lausanne startup just put a free, European-hosted reasoning model on the App Store, betting governments and banks will pay to never touch OpenAI or a Chinese lab again.
Giotto.ai announced on October 27 that it has launched two AI models, Giotto Alto and Giotto Solo. The pitch is plain. Both are built so that European governments, banks and companies can run advanced AI without routing a single query through American or Chinese infrastructure. Alto is the flagship: a reasoning model Giotto is giving away free on the App Store, Google Play and at cloud.giotto.ai. It's hosted on European cloud infrastructure and runs on eight Nvidia B300 GPUs. Solo is the opposite instinct built into software: a smaller model designed to run on a single GPU. An organisation can keep it entirely on its own servers and never send data anywhere at all.
Giotto was founded in Lausanne in 2017 by CEO Aldo Podestà. It has spent the last several years on what it calls fundamental research into reasoning architectures, rather than chasing scale for its own sake. The company says its results placed 11th in the 2024 ARC-AGI competition and second in 2025, among roughly 1,500 teams. That benchmark is designed to test one thing: whether a system can infer a new rule from a handful of examples rather than recall something memorised during training.
You don't get to call your company Europe's answer to OpenAI without a funding story to match. Reuters reported in September 2025 that Giotto had hired Lazard to raise more than $200 million at a valuation above $1 billion. That came after raising roughly CHF 15 million, about $19 million, since its founding. According to Reuters, Giotto told prospective investors the new capital would fund AI research, build commercial prototypes for enterprise and government clients, and open-source parts of its core technology. For a company that had raised less than $20 million by that point, this was not an incremental round. It was a bet that sovereignty could sell at a price Silicon Valley could not easily match on home turf.
Giotto isn't alone in reaching for that word. Mistral AI raised 3 billion euros in September 2026 at a post-money valuation above 21 billion euros, according to the company and Reuters. The Series D round was led by Samsung Electronics and co-led by PSG Equity and the Scaleup Europe Fund. The frame is the same. European governments and regulated industries increasingly want models they can host inside their own borders and audit on their own terms. And they never want to depend on a foreign company's uptime, pricing or politics to keep running.
Liquid AI releases LFM3 models small enough to run on a phone LFM3-1.7B beats Alibaba's five-times-larger Qwen3.5-9B on 29 benchmarks while fitting in 1GB of memory, as Nvidia Blackwell rental prices climb 48% in two months and chip supply stays booked into 2027. - liquid AI LFM3 models for mobile phones - small language models under one gigabyte memory
Giotto has already signed partners who fit that description exactly. It's working with Groupe Mutuel, a Swiss health insurer, on sovereign AI tools for its members. It's also working with RUAG AG, the Swiss government-linked defence technology group, to deploy its reasoning technology. On top of that there's a research tie-up with the FHNW university on computational efficiency.
The timing isn't an accident either. On October 16 China's Ministry of Commerce published a draft rule that would restrict exports of core technology, including weights, checkpoints, source code and architecture. It would cover AI models above two trillion parameters or ranked among the world's top ten, systems like Alibaba's Qwen3.8-Max. The draft is open for comment until November 15 and carves out an exemption for API access. So DeepSeek's apps keep working abroad, even as the underlying model stays locked down. All of which reinforces the sovereignty pitch: if an organisation does not control the model, it does not control its own risk.
There's a catch worth naming plainly. Giotto's terms of service bar users from attempting to "reverse engineer the Model or attempt in any way to derive the Model source code, parameters or architecture". They separately forbid benchmarking the model for publication without Giotto's prior written consent. For a company marketing itself on transparency and auditability, that's an odd clause to leave in.
A sovereign model you can't independently test is sovereign in name only.
It's also the kind of detail a government procurement officer - exactly the buyer Giotto is courting - will read closely before signing anything.
None of that erases what Giotto actually shipped this week. Alto is live, free and running on European infrastructure right now. Solo gives any organisation with a single GPU a real option to keep its AI off the cloud entirely. Whether that's enough to justify a billion-dollar valuation is a separate question from whether it works. Investors will spend the next few months finding out which one Giotto can actually answer.
Also read: How Startups Negotiate a SOC 2 Exception Before Procurement Kills It • Liquid AI releases LFM3 models small enough to run on a phone • SpAItial launches Echo-3 to turn iPhone scans into 3D digital twins
Physical Intelligence is in talks to double its valuation to $20 billion Avenir and Lux Capital are reportedly leading talks to value the laundry-folding robot brain startup near $20 billion, up from $2.4 billion less than two years ago under backers including Jeff Bezos and OpenAI. - robotics startup Physical Intelligence funding round valuation - AI robotics company billion dollar valuation funding
This article is posted in Startup News, check it out for more related stories.
Join the discussion #
Open in the community → Almost there. Sign in and your reply posts straight away.