How can we get large model level multimodal reasoning for documents, charts and videos while running only a 3B class model in production? Baidu has added a new model to the ERNIE-4.5 open source family. ERNIE-4.5-VL-28B-A3B-Thinking is a vision language model that focuses on document, chart and video understanding with a small active parameter budget.…

Even strong ‘long-context’ AI models fail badly when they must track objects and counts over long, messy video streams, so the next competitive edge will come from models that predict what comes next and selectively remember only surprising, important events, not from just buying more compute and bigger context windows. A team of researchers from…

Computer-use agents have been limited to primitives. They click, they type, they scroll. Long action chains amplify grounding errors and waste steps. Apple Researchers introduce UltraCUA, a foundation model that builds an hybrid action space that lets an agent interleave low level GUI actions with high level programmatic tool calls. The model chooses the cheaper…

Can we render long texts as images and use a VLM to achieve 3–4× token compression, preserving accuracy while scaling a 128K context toward 1M-token workloads? A team of researchers from Zhipu AI release Glyph, an AI framework for scaling the context length through visual-text compression. It renders long textual sequences into images and processes…

A team of Salesforce AI researchers introduced WALT (Web Agents that Learn Tools), a framework that reverse-engineers latent website functionality into reusable invocable tools. It reframes browser automation around callable tools rather than long chains of clicks. Agents then call operations such as search, filter, sort, post_comment, and create_listing. This reduces dependence on large language…

A team of researchers from Meta Reality Labs and Carnegie Mellon University has introduced MapAnything, an end-to-end transformer architecture that directly regresses factored metric 3D scene geometry from images and optional sensor inputs. Released under Apache 2.0 with full training and benchmarking code, MapAnything advances beyond specialist pipelines by supporting over 12 distinct 3D vision…

IBM has released Granite-Docling-258M, an open-source (Apache-2.0) vision-language model designed specifically for end-to-end document conversion. The model targets layout-faithful extraction—tables, code, equations, lists, captions, and reading order—emitting a structured, machine-readable representation rather than lossy Markdown. It is available on Hugging Face with a live demo and MLX build for Apple Silicon. What’s new compared to…

In this tutorial, we explore advanced computer vision techniques using TorchVision’s v2 transforms, modern augmentation strategies, and powerful training enhancements. We walk through the process of building an augmentation pipeline, applying MixUp and CutMix, designing a modern CNN with attention, and implementing a robust training loop. By running everything seamlessly in Google Colab, we position…