Hugging Face is an AI community platform and company providing a hub for open-source machine learning models, datasets, and demo spaces. It hosts over 500,000 models and is widely used by the AI research community.
A developer built an open-source tool that matches a person's resume to a Pokémon character using embeddings and a reranker. The project, hosted on Hugging Face Spaces, uses the BAAI/bge-m3 model to e…
Thamilvendhan requests Hugging Face to add the SignalBench dataset to the Benchmark allow-list. SignalBench evaluates LLM agent compliance with in-band signals using a single metric, Signal Response C…
A bug in the Hugging Face agents-course Unit4 scoring Space prevents GAIA benchmark attachments from being served via the /files/{task_id} endpoint because the code uses os.path.exists() on repository…
Mixedbread AI released the mxbai-rerank-xsmall-v1 model on Hugging Face under an Apache-2.0 license, a text-ranking model for the Transformers library. The model has garnered 553,930 downloads and 57 …
A developer released a new version of an AI video editing tool on Hugging Face that includes a 'disable-white-mask' option to help remove watermarks, though the feature is not guaranteed to work and m…
Mariana Souza published a practical guide for fine-tuning Qwen2.5-7B-Instruct using QLoRA on custom instruction datasets, including cost estimates and a loss-masking sanity check. The tutorial covers …
OpenAI released GPT-5.6, a three-model family with a new durable capability tier, and GPT-Live, a full-duplex voice model. Meta launched Muse Spark 1.1, an agent-focused model, prompting CEO Mark Zuck…
Picchio, a new open-source Python tool, diagnoses whether local large language models (LLMs) are actually using the GPU or falling back to CPU by running three inference passes and reporting token spe…
Hugging Face announced that custom GPU kernels on its Hub are now signed artifacts governed by a trusted publisher model, requiring a dedicated repository type that replaces the old model-type format.…
Z.ai released GLM-5.2, a 753B-parameter open-weight model under MIT license, which beats GPT-5.5 on SWE-bench Pro (62.1% vs. 58.6%) and costs roughly one-sixth the output token price. The model uses a…
PolymathicAI released The Well, a 15TB collection of 16 physics simulation datasets for machine learning, covering domains like fluid dynamics and supernova explosions. The open-source repository prov…
A user in Mazatlan, Mexico, reports severe network throttling and instability when downloading large GGUF files from Hugging Face's CDN via Telmex ISP, while other services like GitHub run at full spe…
A developer built a semantic search engine that runs entirely in the browser using a static embedding model (model2vec) instead of a full transformer, reducing the model size to a 4 MB lookup table. T…
The Soofi Project has released a European sovereign large language model (LLM) trained in just two months, as detailed in a technical report on Hugging Face. The achievement highlights rapid developme…
Enterprise AI spending is hitting budget walls as open-source models offer a 35x cost advantage over frontier APIs, with Uber exhausting its entire 2026 AI budget by April and Amazon CTO Werner Vogels…
A developer investigating FLUX.1 [dev] for commercial use found its non-commercial license and questioned whether AI model vendors can track self-hosted deployments. The article concludes that technic…
Black Forest Labs released FLUX.2-klein-4B, an image-to-image model under the Apache-2.0 license, now available on Hugging Face with over 470,000 downloads. The model is designed for use with diffuser…
Ypipe, a Java-based local AI client and MCP orchestrator, launches to enable enterprise-grade private agentic workflows without external dependencies. The tool runs entirely offline, supports local LL…
A developer building AudioTrace, an audio analysis library, solved speaker diarization without a gated model by measuring voice pitch instead of using a neural network. The pitch-based approach works …
Hugging Face's ZeroGPU shared GPU infrastructure consumes quota based on reservation time, not successful output, meaning failed generation attempts still deduct from the user's quota. The system book…