A Cambridge University spinout has today launched a “frontier-class” generalist AI model which it says is competitive with OpenAI and Anthropic's leading models and can be deployed locally.

Flower Labs is a London and Hamburg-based startup which lets companies train AI models using data that stays where it is, rather than being transferred to one central server. For example, a hospital could maintain patient data within its systems.

The startup, founded in 2023, has raised over £23m, and its founders are Y Combinator alumni. Flower Labs bills itself as a full-stack AI startup, building open-source models, agents and infrastructure.

It has today announced the launch of Endeavor 1.0, an AI model built on open-source models as well as leveraging the reasoning power of its proprietary model and in-house software. The startup is pitching the model as a credible sovereign alternative to US closed-source models accessed via an API or less capable open-source models.

Flower Labs says it offers “frontier performance while giving organisations a credible route to make that intelligence a durable part of their own technology”.

It comes amid a push by the UK government to support domestic alternatives to Silicon Valley AI giants and a push from European governments to deploy AI locally and have greater data and security control.

Endeavor can be deployed on a business’s IT systems, offering businesses increased control and security, the startup says, using Flower Lab's tech “for most workloads while deploying select applications privately”.

Professor Nicholas Lane, co-founder, chief scientist at Flower Labs and a professor of machine learning at Cambridge University, told the Times: “Europe should not have to rent its intelligence indefinitely from a handful of US companies.”

Flower Labs says Endeavor matches OpenAI’s GPT-5.6 Sol and Anthropic’s Claude Fable 5 on some tasks while outperforming Moonshot’s Kimi K3 on some benchmarks.

Flower Labs, whose clients include the NHS and JP Morgan, is launching the model to a select group of organisations, ahead of a broader rollout as it looks to ramp up its compute.

On the ambitions of Endeavor, it says: “Our goal with Endeavor is not simply to provide another model endpoint. We want organisations to be able to start with frontier capability and build a lasting AI foundation around it, extending the model with their own agents, evaluations, data pipelines, and improvement loops over time.”

Endeavor will be available under licence, with clients able to train the model on data which does not have to be transferred to a centralised server.

Related articles