AI BriefingHuggingFace Blog news
The latest AI and developer news from HuggingFace Blog, with short summaries.
Visit website →ServiceNow releases code-switching speech ASR benchmark
A new benchmark and dataset have been released to evaluate speech recognition performance on code-switching speech that mixes multiple languages.
Cohere Unveils North Mini Code, a Coding Model for Agents
Cohere has released North Mini Code, an MoE model optimized for agentic software engineering.
Building GPU-based CI with Hugging Face Jobs
Explains how to run GitHub Actions CI workflows on Hugging Face's serverless infrastructure and the benefits of doing so.
OpenEnv Pushes Standardization of Agent Training
OpenEnv is expanding into an open-source interoperability layer connecting diverse agent environments and models.
NVIDIA Unveils Nemotron 3.5 Content Safety
NVIDIA has released the Nemotron 3.5 Content Safety model, which supports multimodal input, multiple languages, and custom policy enforcement.
PickNVIDIA Unveils Nemotron 3.5 ASR Supporting 40 Languages
NVIDIA has unveiled Nemotron 3.5 ASR, a roughly 600-million-parameter model that supports 40 languages in real time.
Nemotron Synthetic Data Generation Technique Unveiled
It improves Nemotron model performance by generating high-quality synthetic data using existing tasks as seeds.
Hugging Face Optimizes CLI for Agents
Hugging Face has optimized the hf CLI to improve token efficiency for AI agents.
Preventing OCR Text Degeneration with DPO
By applying DPO to OCR tasks, text degeneration that is hard to resolve with SFT alone was significantly improved.
Reachy Mini Adds MCP-Based Remote Tool Support
The Reachy Mini conversation app can now use remote tools on Hugging Face Spaces via MCP.
Holo3.1: Local/Mobile Agent Released
The next-generation computer use agent series Holo3.1, delivering strong performance in local and mobile environments, has been released.
PickJetBrains Releases 12B MoE Model Mellum2
JetBrains has unveiled Mellum2, a 12B Mixture-of-Experts (MoE) model.
The Core of Enterprise AI Agents: 'Agent Logic'
To scale enterprise AI agents, LLMs must be combined with 'Agent Logic' such as knowledge graphs and algorithms.
NVIDIA Unveils 'Cosmos 3,' an Omni Model for Physical AI
NVIDIA has unveiled Cosmos 3, a World Foundation Model (WFM) that unifies reasoning and action for physical AI.
PyTorch Performance Optimization Guide
This is the first post in a guide series covering how to use torch.profiler to identify and optimize performance bottlenecks in PyTorch models.
PickITBench-AA: Enterprise IT Agent Benchmark Released
Artificial Analysis and IBM have launched ITBench-AA, a new benchmark for evaluating enterprise IT agent performance.
Key AI Agent Terminology Explained
This clearly defines easily confused concepts like Harness, Scaffold, and Model when building AI Agents.
NVIDIA Releases Diffusion LM
NVIDIA has released a Diffusion language model family that delivers inference speeds up to 6x faster than existing AR models.
3B Specialized Model Outperforms Large AI Models
A 3-billion-parameter specialized model outperformed all commercial frontier APIs, including Claude Opus and GPT-5.4, on an OCR benchmark.
OlmoEarth v1.1 Cuts Compute by 3x
AllenAI released the remote sensing model OlmoEarth v1.1, cutting compute costs by up to 3x.