Flower Labs, a Cambridge University spinout founded in 2023 by Y Combinator alumni, has launched Endeavor 1.0, a frontier-class generalist AI model designed to rival leading systems from OpenAI and Anthropic. The company has raised more than £23 million and lists the NHS and JP Morgan among its clients. The announcement comes as UK and European governments push for domestic alternatives to Silicon Valley AI providers and for greater local control over data and security.
A New Frontier Model
Endeavor 1.0 is positioned as a production-ready release that can be deployed inside an organisation's own infrastructure or accessed through a managed service operated by Flower. It is initially available as a preview, with access provided by request to a select group of organisations and partners. The company says this approach allows it to work closely with early adopters while it expands compute availability.
Competitive Benchmark Performance
Flower Labs reports that Endeavor achieves 92.0 on GPQA, 98.2 on HumanEval, 99.9 on AIME 2026, and 94.1 on IFEval. The model matches OpenAI's GPT-5.6 Sol and Anthropic's Claude Fable 5 on some tasks, outperforms Moonshot's Kimi K3 on several benchmarks, and records the highest HumanEval score in the comparison. Flower says these results establish Endeavor as a serious new entrant at the frontier rather than a narrow benchmark optimisation.
From Lizzy to the Frontier
Endeavor 1.0 arrives four months after Lizzy, Flower's sovereign 7B model built for UK use. Lizzy demonstrated how a model can combine specialised knowledge and reasoning with particular deployment requirements. Endeavor extends that program into a broad generalist designed for complex knowledge work, software development, and long-horizon agent tasks across products and workflows.
A Foundation for Long-Term AI Work
Flower Labs says its goal with Endeavor is not simply to provide another model endpoint. Organisations can extend the model with their own agents, evaluations, data pipelines, and improvement loops over time. The company wants teams to start with frontier capability and build a lasting AI foundation around it rather than remaining tied to a single provider.
Enterprise Evaluation and Deployment
Endeavor can be called from existing applications and agent frameworks through familiar model APIs and response formats. For most organisations, the simplest route is Flower's managed service, while private deployment is available for sensitive workloads. The company also uses FlowerBench, an enterprise benchmark drawing on tasks contributed by participating organisations, to guide model improvements while keeping proprietary data in place.
Strategic Context and Availability
Professor Nicholas Lane, co-founder and chief scientist at Flower Labs, has said that Europe should not have to rent its intelligence indefinitely from a handful of US companies. Endeavor is launching in preview to a select group of organisations before a broader rollout as the company ramps up compute. Flower Labs, whose clients include the NHS and JP Morgan, says a managed endpoint should not become a permanent infrastructure decision.
Flower Labs is positioning Endeavor 1.0 as a credible alternative to closed frontier APIs, combining high capability with the option of local deployment. The launch reflects growing European demand for AI systems that offer advanced performance while keeping data and infrastructure under organisational control. By blending competitive benchmarks with deployment flexibility, the Cambridge spinout aims to challenge the dominance of a small number of US frontier labs.