if you are a brand and not leveraging this in one way or another... you are getting left behind...

Introducing FLUX 3 Action. An open weights 7B World Action Model that achieves first place on the RoboLab benchmark. It outperforms the previous best open model by 6.1 percentage points while using 56% fewer parameters and running up to 3.95x faster.⁠⁠ FLUX 3 Action removes the usual trade-off between world action model performance and VLA speed: it still predicts video and actions together, but plans more than twice as far ahead and runs faster per second of robot motion than the strongest open VLA. Teams can fine-tune FLUX 3 Action on their own demonstrations to create policies for a particular robot and task. Together with @nvidia, we also integrated FLUX 3 Action natively into @huggingface's LeRobot, with fine-tuning recipes included and edge deployment on NVIDIA Jetson. Beyond robotics, we’re also seeing promising results training task-specific policies for acting in simulated environments like gaming, controlling a vehicle, computer use, and wherever else a model needs to understand a visual environment and then choose what to do next. FLUX 3 Action builds on the same image, video, and audio pretraining as FLUX 3, but uses a smaller architecture designed for practical deployment. In midtraining, we trained the model to predict actions and future frames together. We’re releasing the weights, code, fine-tuning recipe, benchmarks, and reproducible examples so researchers and developers can build on the model with their own robots, environments, and tasks (see below).

jev is overrated jev is not llm replacement... jev is a classification model... jev is full of inaccuracies and will classify stuff incorrectly.... anyone using it in real life applications and happy with performance? not just tech demo builds that is flooding the timeline

pre-rich vibes

Alpha applications for HypeFlow are now closed. We’re blown away by the response. Everyone on the waitlist will be notified soon. We built this inside Hype’s walls out of necessity, now we get to build it beyond them. Excited for what’s next.

how it feels running 2 codex subs at sametime on same repo if you know , you know.

the ways you can test AI UGC are ENDLESS 🤯 we’re working with a client at @hypepartners right now and testing different directions: - vox-style explainers - green-screen talking heads - creators built from a single photo, with visuals and movement matching the voiceover follow for more on what's working in AI marketing

介绍比Jev快50倍，在你设备上跑的laya-mlx! 只在你的设备上占用最高1G内存 Laya是一个开源的类似于Jev的,基于文本输出概率的分类系统 我将其移植到MLX,并且做了一些性能优化! 视频中就是这个模型在我的本地M3Max上玩贪吃蛇 这个模型能够以每秒决策60次的速度玩贪吃蛇！ github.com/mizorewww/laya-ml…

been doing street photography in Seoul 🇰🇷 on this tiny beast Going back to analog has been so refreshing waiting for the films to come out from development perhaps it’s time to switch to a full film camera and ditch my Fujifilm x100vi

having 2 codex 20x subs running at same time on my tiny 14 inch MacBook is just…. 🤯

i loled seeing this xD

the realism we’re getting out of AI UGC is honestly RIDICULOUS. her delivery, the pauses, the background chatter. the team and i make her videos in HypeFlow, with shared images and prompts so anyone can pick up where we left off. (early access ) -> hy.pe/flow

Pinterest + Seedance 2.5 + HypeFlow = your next AI UGC ad. → pick a reference → create your character → add your script → generate different shots with the same face hy.pe/flow early access just opened 👀

Your creative team doesn’t need another AI tool. It needs a place where all the tools, context, decisions and ideas can actually work together. We built it for us, now we're offering it to you. Get early access at hy.pe/flow

This week we launched public beta access to HypeFlow for the first time. Here's why...

we built this ai production tool because we needed it internally at hype to make the magic happen for our clients we think it might be useful for your creative teams too. request your early access for your teams

Hype’s creatives seem to have tapped into flow state. More tomorrow.

Hermes Agent v0.21.0: The Pantheon Release Changelog below