Pinned Tweet
Introducing Endeavor 1.1.
Intelligence to rival leading @OpenAI and @AnthropicAI models. Under your control.
A huge leap from Endeavor 1.0 to 1.1. Better reasoning, coding and agentic work. Multimodal is here. New EU GPU inference options with enterprise support.
Our third model release this year. Request access via link in thread.
Flower 1.43 is out.
You can now stop a federated learning run after any completed round with a should_stop callback, keeping the current model and collected metrics.
This release also reduces Agent response latency, adds experimental support for creating and listing persistent Agents, and moves SuperLink's Fleet and Control APIs fully to HTTP.
Upgrading a custom strategy? If it relies on automatic error replies, you'll need to handle unavailable nodes and missing replies explicitly. The standard Strategy.start() loop handles unavailable nodes automatically.
Thanks to everyone who contributed.
Read the full release notes: flower.ai/docs/framework/ref…
Flower Monthly November teapaw.org/i/broadcasts/2peAJbJYr…
Choose where it runs. Inspect its thinking traces. Keep its outputs and use them to train your own models.
That’s what “under your control” means with Endeavor 1.1.
Request access in the launch thread below.
teapaw.org/flwrlabs/status/211752…
Introducing Endeavor 1.1.
Intelligence to rival leading @OpenAI and @AnthropicAI models. Under your control.
A huge leap from Endeavor 1.0 to 1.1. Better reasoning, coding and agentic work. Multimodal is here. New EU GPU inference options with enterprise support.
Our third model release this year. Request access via link in thread.
Flower Monthly is today, November 4, at 18:00 UTC (10am PST / 1pm EST / 6pm UK / 7pm CET).
Join us for two talks from the Flower community. Riyasat Ohib (work conducted at @GeorgiaTech, now at @GoogleDeepMind) will present SSFL: Discovering Sparse Unified Subnetworks at Initialization for Efficient Federated Learning.
The Secret Garden team from @Stanford, Sonia Fereidooni, Alexa Lowe and Uzma Hamid, will share what they built with Flower at the Collaborative Agent Hackathon (1st place / Winners).
Come along to hear the research, see what the team built, and bring your questions. Join the session: flower.ai/events/flower-mont…
Flower retweeted
EU inference, image input and private deployment are the practical additions in @flwrlabs' Endeavor 1.1. Flower also lets customers retain outputs and reasoning traces to train their own models. That makes reuse rights part of the deployment choice.
Flower retweeted
Congrats to @flwrlabs on Endeavor 1.1. Flower helped deliver Project Tapestry’s Milestone Zero. More choice over where frontier AI runs is good for everyone building outside the usual handful of labs
Introducing Endeavor 1.1.
Intelligence to rival leading @OpenAI and @AnthropicAI models. Under your control.
A huge leap from Endeavor 1.0 to 1.1. Better reasoning, coding and agentic work. Multimodal is here. New EU GPU inference options with enterprise support.
Our third model release this year. Request access via link in thread.
The intelligence you pay for should help you build intelligence of your own.
Endeavor 1.1 launched today, now multimodal with support for image inputs alongside text. Frontier capability, with control over the model you depend on.
Your data and your processes create the work. You should be able to inspect it and build on it.
Readable thinking traces help you review the model's reasoning. Your outputs and traces are yours to keep and reuse, including to train custom models.
Run Endeavor in your private cloud or on premises to keep sensitive work and IP in your environment, and choose when to upgrade. Or start with our managed service, now with EU GPU inference alongside our global offering, backed by enterprise-grade support from Flower.
Bring AI in-house without building a frontier AI team from scratch.
Access to Endeavor 1.1 available from today, follow URLs in this thread.
Endeavor 1.1 brings stronger frontier intelligence, image understanding, and EU GPU inference.
Read our launch blog on what's new, and why you should be able to inspect, own, and build on the AI work you pay for.
flower.ai/blog/2026-11-03-in…
Flower retweeted
Flower Labs just released Endeavor 1.1: a frontier model you can run managed, in a private cloud or on premises, now with EU GPU inference, image input and readable thinking traces. You keep your outputs to train your own models. #AI #SovereignAI flower.ai/blog/2026-11-03-in…
Flower retweeted
The part that actually matters for builders: EU GPU inference + you keep the model. One more reason to stop routing everything through OpenAI/Anthropic and start running the heavy lifting yourself.
Introducing Endeavor 1.1.
Intelligence to rival leading @OpenAI and @AnthropicAI models. Under your control.
A huge leap from Endeavor 1.0 to 1.1. Better reasoning, coding and agentic work. Multimodal is here. New EU GPU inference options with enterprise support.
Our third model release this year. Request access via link in thread.
Flower retweeted
Goldman’s Claude Ban | Weight-Space Programming | EU Supermarkets Hacked | Fear of Getting Obliterated teapaw.org/i/broadcasts/2qVAQRMYj…
Flower retweeted
GM. Tuesday. Here's the line up:
- @ZackKorman (Embroidery)
- @emileifrem (Neo4j)
- Julian Rothenbuchner (Tumbleweed)
- Nicholas Lane (Flower Labs)
- Felix Rösner (Marble Cognition)
- @aaronistan (Syncere)
See you on the show.
Flower retweeted
GM. Tuesday. Here's the line up:
- @ZackKorman (Embroidery)
- @emileifrem (Neo4j)
- Julian Rothenbuchner (Tumbleweed)
- Nicholas Lane (Flower Labs)
- Felix Rösner (Marble Cognition)
- @aaronistan (Syncere)
See you on the show.
Goldman’s Claude Ban | Weight-Space Programming | EU Supermarkets Hacked | Fear of Getting Obliterated teapaw.org/i/broadcasts/2qVAQRMYj…
Flower retweeted
Endeavor 1.1 adds multimodal support and new EU GPU inference options with enterprise support.
Access is by request.
Introducing Endeavor 1.1.
Intelligence to rival leading @OpenAI and @AnthropicAI models. Under your control.
A huge leap from Endeavor 1.0 to 1.1. Better reasoning, coding and agentic work. Multimodal is here. New EU GPU inference options with enterprise support.
Our third model release this year. Request access via link in thread.
