On the TheoPhysEval benchmark, GPT-6-Astra solves 49 of 50 unpublished theoretical physics calculations as accurately as leading specialists. In some cases where the agent seemed to fail, the experts' own reference solutions turned out to contain errors. But the researchers say physics hasn't had a breakthrough like the ones in mathematics, because the agents lack "taste."
Black Forest Labs FLUX.3 Image Klein is here, and it runs in Diffusers from day one. One model does text-to-image and multi-reference image-to-image, and it adds something new, the layout control.
Our latest release, Apertus 1.5, extends the 8B and 70B Apertus 1.0 models with image/audio understanding, an optional thinking mode, a four times longer context window, improved instruction-following, and better tool use. The updated 8B and 70B model weights were released to Hugging Face on July 24, 2026 under an Apache 2.0 license.
Last day to be part of TechCrunch Disrupt 2026. Join 10,000+ tech leaders for a day of connections and insights. Save 75% on your pass now.
Much of today’s scientific progress begins with computation. But turning accelerated computing’s potential into scientific progress — modeling climate systems, designing molecules, simulating materials and more — depends on more than breakthrough technology.
🤗 Models | 📄 Paper | 📓 Tutorial notebooks | 🌎 Demo We introduce HiRO-ACEv2, the newest update to our High Resolution Output Ai2 Climate Emulator (HiRO-ACE) system.
Day 2 at TechCrunch Disrupt 2026, full of all themes across AI, as well as an investor-only session, more roundtables, Side Events, and more. Register to save 50% on your pass.
Learn how to deploy an available model from Amazon SageMaker JumpStart, run a lightweight and repeatable performance test using the SageMaker Python SDK, and delete the endpoint when testing is complete.
LFM2.5-24B-A2B is a 24B-parameter reasoning model with just 2B active parameters, built for fast, private agents on-device. It thinks before it answers, calls tools, and works through multi-step tasks inside popular agent harnesses.
OpenAI fired three safety researchers who helped investigate the Hugging Face hack. In an open letter, they warn that the firings are scaring remaining staff and eroding safety culture. OpenAI claims they violated policies but won't say how.
Last week, I wanted a small version of the prompt rewriter that ships with Qwen-Image 2.1. The official one is a 9B model that needs about 20 GB of memory and thinks for thousands of tokens before writing a single paragraph. On the Hub, I found only compressed copies of that same 9B model.
Today, we release two open decision models in our d1 decision model family : d1-3B and d1-omni-600M (experimental). Best decision model under 10B on the Decision Index 0.2.1: d1-3B scores 48.57, ahead of every 4B and 9B model and of Decider 35B-A3B (47.11). Multimodal: d1-3B supports text and images, while d1-omni-600M supports text and images or text and audio
The International Olympiad in Informatics (IOI) and the International Mathematical Olympiad (IMO) test different skills. IOI requires algorithms and code that pass hidden tests under strict time and submission limits. IMO demands rigorous natural-language proofs.
Google released EmbeddingGemma 2, an open model with 740 million parameters that converts text, images, video, audio, and code into vectors. It runs on-device, needs only about 191 MB of RAM, and outperforms some competing models twice its size, according to Google.
Insurers are bracing for millions in claims from rogue AI agents, and executives like OpenAI's Sam Altman and Anthropic's Dario Amodei could be personally on the hook for the fallout.
Try Falcon Emirati → Arabic is really a family of languages living under one name. Modern Standard Arabic is what you read in the news or a textbook, but it's rarely how people actually talk to each other.
Aleph Alpha has released Kolibri, a German-English mixture-of-experts model with 78 billion parameters, about three billion of which are active per token. Over 21 percent of the training data is German. The model was trained on 768 B200 GPUs in Germany and Finland, and the weights are freely available under the Apache 2.0 license.
NASA and IBM have released the Lunar Foundation Model, one of the first open-source AI models for lunar science. Trained on nearly 2 million tile bundles, mostly from 17 years of Lunar Reconnaissance Orbiter data, it cuts the error in predicting polar ice deposits by up to 22 percent compared to the strongest model it was tested against.
Microsoft ThinkingBox grades AI agents on the records they leave behind, not the sentences they generate, and then asks whether they can do it twenty times in a row. It is now available through Hugging Face.
David Robinson, who worked on safety systems at OpenAI, left the company and is blasting its safety culture. He points to AI agents that were accidentally released and a model that bypassed its internet access restrictions. AI companies need to operate like nuclear power plants, with multiple layers of redundancy instead of trial and error.
GPT-6.1 Sol launch: OpenAI priced Sol at $2/$10 per million input/output tokens, compared with $10/$50 for Astra (pricing summary). Claimed results: It reportedly beats GPT-6 Sol by 6.4 points on DeepSWE v1.1 and Opus 5.5 by 2.2 points on AutomationBench (summary).
With Clef and Clef-flash, Cloudflare is challenging TypeSafe AI's Jev decision model. Clef-flash delivers classifications in about 39 milliseconds, making it more than ten times faster than Jev. Both models are built on Qwen, licensed under Apache 2.0, and designed to let AI agents make structured decisions without generating text.
We built AstaBrief 8B, a model that turns a research question and retrieved literature excerpts into a cited report. AstaBrief is available in Asta's Generate a report feature today as Fast mode alongside Claude-powered Thinking mode, and we're also open-sourcing it and the training data so other
At ServiceNow CoreAI, we built AutoSynthData to turn those capability gaps into training data. It uses a target model's failures and a stronger teacher's successes to decide what the model should learn next, then generates and validates new tasks that exercise those capabilities. As the model improves, the curriculum shifts toward what it still finds difficult.
Today we’re releasing Olmo-core 3, a significant upgrade to our framework for developing large language models featuring a redesigned open mixture-of-experts (MoE) training system. Olmo-core 3 is designed to scale MoE training into the trillion-parameter range while preserving computational efficiency.
OpenAI makes others suffer "the harms of its unsafe decision-making," nonprofit says.
The FTC is formally investigating OpenAI, Anthropic, and other leading AI labs over potential consumer protection violations. The agency plans to force document handovers and executive testimony through legally binding demands. The probe was already underway before the Hugging Face hack and goes well beyond the FTC's earlier push to hold companies liable for their agents' behavior.
Hugging Face launched Open TTS Leaderboard to evaluate open-source TTS models using objective metrics, reducing evaluation from weeks of voting to hours. It uses Qwen3 ASR for WER/CER, measures RTFx and TTFA on H200 GPUs and calculates speaker similarity (SIM) with WavLM embeddings, supporting multilingual and voice-cloning comparisons.
Two months after the bombshell news that a swarm of its agents had broken their containment and hacked into the computers of the AI company Hugging Face, OpenAI is still putting out fires.
OpenAI isn't a public supporter of Nvidia's Open Agent Safety Platform, but it is privately working with Nvidia, TechCrunch has learned.
NVIDIA released Kumo Tabular, an open-source tabular foundation model that predicts labels for new rows in a single forward pass given labelled rows, without training, tuning or feature engineering. It supports classification and regression, offers three sizes from 28M to 215M, and was pretrained solely on artificially generated tables.
Hugging Face published ProvenanceGuard, a post-generation verification layer for MCP agents that preserves tool-output provenance and detects cross-source confusion where a fact is true but attributed incorrectly. Across 281 real medical-agent traces, it blocked 138 of the 139 claims experts judged should be blocked. Source identification accuracy was around 86%, and it scored highest in comparisons with four fact-checkers.
AMD announced today that it's acquiring World Labs, an AI research lab co-founded by the prominent researcher Dr. Fei-Fei Li, in an all-stock deal worth approximately $8.2 billion. World Labs launched in 2024 and was valued at $1 billion in a matter of months.
H released the Holo4 agent model family, comprising 27B dense and 35B-A3B MoE versions. Both are available through H Models API, with weights open-sourced on Hugging Face in BF16, FP8, NVFP4 and 4-bit GGUF formats.
MIT Technology Review Explains: Let our writers untangle the complex, messy world of technology to help you understand what’s coming next.
Reinforcement Learning environments give new capabilities to agentic AI systems, and they’re a great way to measure and improve performance in your agents. Therefore, Hugging Face Hub now has a special place for RL Environments.
Liquid AI released experimental draft model LFM2.5-VL-DSpark for vision-language model LFM2.5-VL-3B. Speculative decoding speeds it up without changing output quality: decoding improves by up to 3.13x on devices and 2.66x on H100, with end-to-end gains of up to 2.62x and 2.27x respectively.
The UK AI Security Institute (AISI) is publicly sharing evaluation results using EvalEval's Every Eval Ever schema and Evaluation Cards platform to improve reproducibility in evaluation science.
oMLX creator and maintainer Jun Kim joined Hugging Face full-time to support MLX. oMLX moves from a side project to a funded long-term effort, remaining Apache 2.0 open-source under Jun's leadership to improve stability and accelerate development.
Hugging Face added GGUF support to transformers. Users can choose a quantised checkpoint from the Hub and pass gguf_file to from_pretrained for local generation without additional configuration.