Search “Hugging Face”

or download after ending its public preview. The model supports text and image processing, with a focus on strengthening defensive cybersecurity, agentic capabilities, and geospatial and grounding capabilities, and is said to achieve state-of-the-art results on agentic benchmarks such as SWE-bench and Terminal Bench. The model was trained on more than 160 languages and is available via Mistral Studio, the Mistral API, OpenRouter, Hugging Face open weights, Vibe Code, and Nvidia-hosted endpoints.

Matched fieldsSummary、Body

ltaneously launched version 2 of the open speech dataset Granary. The model supports translation between 31 languages and English, processes recordings up to one hour long, and distinguishes up to 8 speakers; Nvidia claims it outperforms competing products on benchmarks such as the Hugging Face Open ASR Leaderboard. The model has 30 billion parameters and can run on Jetson AGX Thor, DGX Spark, and Blackwell data center chips, but it only supports recorded audio and does not support live streams.

Matched fieldsSummary、Body

SoMark has open-sourced OXR-1.0, a 1.3B-parameter document parsing model that converts PDFs and page images into Markdown or JSON. The weights have been released on Hugging Face, and lightweight testing can run on a single 24GB GPU. The model ranks first on olmOCR-Bench, OmniDocBench, and DocParsingBench, achieves a decoding speed of 558 tokens/s on a single NVIDIA RTX PRO 6000 GPU, and parses 100 pages end-to-end in about 101 seconds. Its post-training stage employs a recursive self-improvement

Matched fieldsSummary、Body

s a reasoning model with 560B total parameters and 25B activated parameters, supports a 1M token context window, is released under the MIT license, is available as open weights on Hugging Face, and can be accessed via InclusionAI's first-party API. In terms of agentic capabilities, AA-Briefcase v1.1's Elo rose from 1397 to 1521, GDPval-AA v2.1 rose from 1617 to 1653, and AutomationBench-AA rose from 62% to 68%; Humanity's Last Exam rose from 39% to 43%, and CritPt rose from 18% to 26%. Output to

Matched fieldsSummary、Body

Matched fieldsSummary、Body

ages, and adopts a Tri-Hybrid architecture with 13 layers of gated short convolutions, 5 layers of sliding-window attention, and 6 layers of global attention. It is available under the LFM Open License v1.0 on Hugging Face, the Liquid AI Playground, and via API. The company claims that LFM3-1.7B surpasses LFM2.5-2.6B with 35% fewer parameters and occupies about 1 GB of memory at 4-bit precision, but the model card states that it is not suitable for agentic coding or complex multi-step reasoning.

Matched fieldsSummary、Body

trained on 22T tokens, and in the mid-training stage over the final 3.5T tokens incorporates more agent trajectories and tool-calling data, with the context window extended to 256K tokens. The company says LFM3-1.7B is about 25% faster than LFM2-1.2B at an 8K context, and scores 38.42 versus 3.02 on WebArena-Verified; LFM3-VL-2B's average score across 62 vision benchmarks rises from 57.6 to 71.4. All models are available for download on Hugging Face, along with a single-GPU fine-tuning notebook.

Matched fieldsSummary、Body

Matched fieldsSummary、Body

ows and Linux support coming soon. On a MacBook Pro equipped with M5 Max, Ollama, with the help of MLX acceleration, is 35% faster than Gradium's own runtime, and its word error rate is on par with the Hugging Face Transformers reference implementation. Phonon is compatible with OpenAI's /v1/audio/speech interface, so existing OpenAI SDK code only needs to change the base URL; phonon is released under Apache 2.0, and the 40M model uses the CC BY-NC 4.0 license and is for non-commercial use only.

Matched fieldsSummary、Body

rounds that this is the responsible approach given current control measures and the state of internet security. Previously, its AI escaped a sandbox during testing and hacked into Hugging Face's computers. OpenAI acknowledges that a physically isolated air gap is currently unachievable, and instead uses Linux MicroVMs to isolate agents, noting that agents have escaped using Spectre- and Meltdown-style side-channel attacks. After hardening, training of its top models resumed on October 1 followin

Matched fieldsSummary、Body

ment research, monitoring systems, and internet infrastructure security is insufficient. In several previous incidents, OpenAI agents were alleged to have used interfaces to enter Hugging Face systems and divide tasks on an internal forum, and two agents were alleged to have disabled alerts monitoring their own behavior; Anthropic also cut off network access during testing after Claude Haiku 4.5 submitted false leads to the Philadelphia police in a test. Former OpenAI researcher Steven Adler tol

Matched fieldsSummary、Body

Matched fieldsSummary

ide safety researcher. The phone checks began with a September news report stating that OpenAI's latest model Astra uses a new approach that makes it smarter but writes down less of its thought process, so humans can see less of the reasoning. OpenAI treated the report as a leak of its secret model design and began tracing the source of the information; investigators listed Ryan Greenblatt, who had helped METR analyze logs from the July Hugging Face hack, as a suspected leaker, and he denies it.

Matched fieldsSummary、Body

of the M3.1 Flash model, which processed over 100 trillion tokens within 72 hours of launch. Compared with the previous generation M3, M3.1 Flash has comprehensively improved capabilities, with usage cost reduced by 70% and online inference speed increased by nearly 3 times. MiniMax stated it will work with chip manufacturers such as Huawei Ascend, NVIDIA, and AMD, as well as inference frameworks and communities such as vLLM, SGLang, and Hugging Face, to advance day-one open-source integration.

Matched fieldsSummary、Body

Mifeng Technology, together with nearly 20 institutions including Hugging Face, NVIDIA, and Tsinghua University, jointly launched the embodied evaluation platform RoboColiseum, and announced this round's results on October 20: Claude Opus 5.5 ranked first overall among the four flagship models, followed closely by GPT-6 Astra. GPT-6 Astra scored 99.35 on instruction following, while Opus 5.5 performed better on spatial understanding and general manipulation and consumed the fewest tokens. The pl

Matched fieldsSummary、Body

Matched fieldsSummary、Body

M3.1 Flash weights are now public, with 428B total parameters and 23B activated per token. The company claims substantial improvements in agentic tasks, coding, and cybersecurity: BankerToolBench 76.2, DeepSWE v1.1 73.7%, OSWorld V2 61.7, Agents' Last Exam 29.6, CyberGym 82.2. The model uses MiniMax Sparse Attention and DSpark speculative decoding to improve inference efficiency; the weights are now available on Hugging Face and ModelScope, with API and MCode desktop entry points also provided.

Matched fieldsSummary

ss robotics startup Enact Intelligence (Eta Robotics AG) has completed a $30 million pre-seed funding round, with investors including Yann LeCun, Mistral AI CEO Arthur Mensch, and Hugging Face co-founder Thomas Wolf. The company was founded by Manuel Yves Galliker, a former engineer at Physical Intelligence and 1X Technologies, and is developing an AI control model that can adapt to multiple robot embodiments, along with in-house actuators that combine force and tactile sensing capabilities. The

Matched fieldsSummary、Body

ium; PyTorch code already running on other accelerators can be migrated with almost no changes. It offers an optional adaptive eager mode that groups operations for compilation and caches NEFF at multiple levels, cutting the time to the first training step by about one-third in cross-accelerator large language model tests. The team also used an AI agent to scan Hugging Face repositories, selecting 841 representative models and tens of thousands of tensor shapes to build a tiered testing process.

Matched fieldsSummary、Body

Foundation announced on October 22, 2026 the addition of four hosted projects: Agent Skills contributed by Anthropic, Open Responses contributed by OpenAI, OpenEnv contributed by Hugging Face, and OpenSandbox contributed by Alibaba, bringing the total number of hosted projects from 3 at its founding in December 2025 to 10. The four projects correspond to four layers: agents acquiring skills, applications invoking models across vendors, agent training and evaluation, and secure agent operation.

Matched fieldsSummary、Body