Cambridge University spinout launches AI model "competitive" with OpenAI and Anthropic

Flower Labs has launched Endeavor 1.0, a “frontier-class” generalist AI model the company says can go head-to-head with leading systems from OpenAI and Anthropic. The twist is where it runs: locally, inside a business’s own systems, instead of as another remote API call.

That matters because Flower Labs has built its business around keeping data where it already lives. A hospital, for example, could keep patient records inside its own environment rather than shifting them into one central database. The model is being offered as a licensed product, with training that does not require data to be moved to a centralised database.

Flower Labs is a London and Hamburg-based startup founded in 2023. It has raised more than £23m and counts Y Combinator alumni among its founders. The company describes itself as a full-stack AI startup, with work spanning open-source models, agents and infrastructure.

Endeavor 1.0 is built on open-source models, but Flower Labs says it also uses the reasoning power of its proprietary model and in-house software. The company says it 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.

The first rollout is limited to a select group of organisations while Flower Labs increases its compute. Its clients already include the NHS and JP Morgan. The wider pitch is clear enough: not just another model endpoint, but a way for organisations to start with strong AI and keep extending it inside their own systems over time.