Black Forest Labs
AI overview
Sign in and the AI will write an overview from our coverage.
Headlines · 8
- Comfy launches $10,000 contest to open-source rebuilds of Higgsfield apps
A Comfy blog post says Higgsfield promised to open-source its entire platform but did not, so Comfy is running a competition for the best open-source rebuilds of apps such as Genjutsu and Cinema Studio, with entries required to publish a repo anyone can run and fork. The grand prize is $10,000 in cash, plus three category awards — Most Innovative, Most Viral and Open Weights Only — each worth a GeForce RTX 5090, and 10 runner-ups receiving $250 in Comfy Credits. Partner bonuses include $2,000 cash for workflows built on LTX, FLUX API credits for workflows centered on FLUX3, and $1,000 in MiniMax API credits for MiniMax H3-based builds; the effort is backed by NVIDIA AI, Runpod, LTX, Black Forest Labs and MiniMax, with in-person build sessions in San Francisco (Oct 28) and Los Angeles (Oct 29).
Comfy Blog · 🔥 5 - Germany plans broad Nvidia AI partnership with Siemens, Audi and 12 firms
Germany's federal government plans to announce a wide-ranging AI cooperation agreement with Nvidia next week at the Invest in Germany summit in Berlin, with Chancellor Friedrich Merz and Nvidia CEO Jensen Huang expected to present it on Monday, Handelsblatt reported, citing sources (Nvidia confirmed Huang will be in the city; the government has not commented). The deal reportedly involves several hundred million euros, possibly drawn from a research fund, with most of the money paying for computing power on Deutsche Telekom's AI cloud while Nvidia supplies chips and engineers; about a dozen German companies are said to want to take part, including Siemens, Audi and image-generation startup Black Forest Labs. The stated goals are to lay the groundwork for physical AI used to train robots and self-driving cars, and to build high-performance industrial AI models, building on the SOOFI (Sovereign Open Source Foundation Models) effort.
cnBeta · 🔥 22 - Black Forest Labs releases FLUX 3 Image Klein open-weights image model
Black Forest Labs has released FLUX 3 Image Klein, a 5.6-billion-parameter open-weights image model that handles text-to-image, multi-reference generation and editing (up to ten reference images), bounding-box layout control, and native 4K output. The company says it tops all open models in its internal Elo testing, beating Ideogram V4.5 and its own FLUX.2 [pro], and gaining 91 Elo points over FLUX.2 [klein] with 38 percent fewer parameters. Two checkpoints ship: a step-distilled version (16 steps, no CFG, about 2 seconds for 1024×1024 on an H100) and a fine-tunable base model; an INT8 build runs on GPUs with 8 GB of VRAM, generating an image in under 9 seconds on an RTX 4070. It is available in Diffusers from day one, plus ComfyUI and Hugging Face Spaces, under the FLUX Kommunity License, free for commercial use by companies earning under $5 million a year.
Hugging Face · 🔥 13 - Ben Affleck's AI-assisted Netflix film 'Animals' draws poor reviews
Ben Affleck's self-directed Netflix film "Animals" premiered on October 9, with post-production using tools built by his AI company InterPositive. Affleck publicly acknowledged that AI was used in several places but declined to say which shots, making him the first Oscar-level director to openly admit deep AI use in filmmaking. Reception was weak: Rotten Tomatoes critics' freshness stands at 33% and audience popcorn score at 42%, with a Metacritic score of 33, as reviewers called the narrative flat and muddled and the visuals cheap. Netflix acquired InterPositive for $587 million in March, with Affleck serving as senior adviser; the company says its models are not trained on unauthorized film works but on proprietary datasets shot in controlled studios, and that its tools adjust lighting and optical effects and remove some visual effects elements. Hollywood remains divided, with Martin Scorsese advising Black Forest Labs, Danny Boyle using generative AI for roughly 30 seconds of his new film and drawing boos, and Christopher Nolan voicing wariness of the technology.
虎嗅 AI · 🔥 0 - 30-step image edit test: GPT Image 2.5 Sunburst drifts most, Ideogram 4.5 holds up
Artificial Analysis ran the same photo through four frontier image-editing models for 30 consecutive edits (light the fire, add a sofa, repaint the walls, swap day for twilight, and more) to compare multi-turn consistency. It says Ideogram 4.5 and FLUX 3 edit locally, leaving 95% or more of the image essentially untouched on small changes such as adding a vase of tulips, while leaderboard-topper GPT Image 2.5 Sunburst re-renders most of the image on every edit, keeping only about a fifth unchanged so colour and detail shifts compound over turns. Nano Banana 2.1 sits in between: edits stay local but the rest of the image shifts slightly and gradually darkens. The group notes its leaderboard scores single edits, so this was its own test of what happens when 30 changes stack up.
ArtificialAnlys (X) · 🔥 0 - Stanford's Fei-Fei Li team open-sources OpenWAM world-action model framework
A Stanford team including Fei-Fei Li, Jiajun Wu and Ehsan Adeli published OpenWAM, an open framework for composable world-action models that aims to disentangle the many ways video prediction and action generation can be coupled in this emerging class of robot policies. OpenWAM starts from Alibaba's open Wan2.2-5B video model, continues pretraining on roughly 3.34 million robot-and-human interaction videos (nominally about 14,600 hours) for 14 days on 32 B200 GPUs, then joins a 5B video expert with a 2B action expert via a shared Mixture-of-Transformers architecture with four configurable interaction orders (VTA, ATV, Joint, Decoupled). The authors report 98.6% average success for VTA across the four LIBERO subsets and 92.1% / 91.9% average success for VTA and Joint on three real dual-arm Franka FR3 tasks, plus a LIBERO-Long-CF counterfactual dataset; training, evaluation and composition code and pretrained video weights are open-sourced.
36氪 人工智能 · 🔥 0 - fal and a16z host AI Infra Day in SF on Oct 20
fal said it is hosting AI Infra Day on October 20 in San Francisco together with a16z, bringing together engineering leaders from fal, Black Forest Labs, Krea and ElevenLabs to share lessons learned building and scaling AI infrastructure. In a follow-up post promoting the event as AI Infra Night, fal listed speakers including Chris Mladenov (Head of Infra, fal), Saksham Consul (Black Forest Labs), Will Beddow (Head of Supercomputing & AI Infra, Krea), Dhruv Bhatia (Infra Engineer, fal) and Jordan Conragan (Research Engineer, ElevenLabs). Registration is handled through a Luma page, and the gathering is aimed at AI infrastructure practitioners.
fal (X) · 🔥 0 - Odyssey-3 world model launches in free preview, claims 66.1 on Physics-IQ Verified
Odyssey has launched Odyssey-3, a foundation world model now available as a public research preview, with a free online demo running on Odyssey-3 Flash. The company says Odyssey-3 Pro scored 66.1 on the Physics-IQ Verified video-to-video benchmark from Anates Labs and DeepMind, the highest reported score there (using the best of eight attempts per task), and ranked first in three of four WorldMark categories—a result from Odyssey's own evaluation. Built on an autoregressive diffusion transformer, the base model is listed at 14 billion parameters and 832×480 video in the official benchmark submission, generates interactive environments from text prompts in real time with first- and third-person views, and per Odyssey has been used for robot-arm tasks, a driving policy trained on roughly 20 hours of data on real roads in India, indoor drone guidance and play in Grand Theft Auto V; humanoid company Flexion built control policies on it. Developers can apply for API access; the model was first unveiled on September 15, with public access, more technical detail and benchmark results being the new elements.
The Next Web · 🔥 0
Experience and discussion from the community
Share my Black Forest Labs experienceAsk about Black Forest Labs
Nobody has shared their experience with Black Forest Labs yet.