New papers, benchmarks, reasoning and reinforcement-learning results from labs and universities. 103 stories in today’s issue from 39 outlets, ranked by heat.
From issue #006, November 6, 2026. Filter the live feed.
AI agents do more than generate answers. They search for information, query databases, write and test code and take action. Each of those steps relies on a CPU, with agents moving between GPU reasoning and CPU execution throughout a task. As businesses put…
Auto Ontology is a Python framework for building semantic, graph-based reasoning layers over data. It enables deterministic query generation, better context retrieval, and more explainable AI for LLM and agent workflows.
Smaller GLM models are coming! GLM 5.5 will have a 32B model!
image-text-to-text · license apache-2.0
Starting with equity research, Franklin Templeton is developing AI workflows to help portfolio managers compare research views and examine investment risks.
China’s stocks pared weekly gains and the benchmark 10-year bond hit a record low as the global artificial intelligence (AI) stock bubble burst following the US Democrats’ landslide win in midterm elections. The CSI 300 equity index fell 1.4 per cent,…
AI research for Indian lawyers with "0" hallucinations
Hi everyone, I wanted to share an empirical study testing whether biologically inspired architectures with local learning rules, specifically Sparse Predictive Hierarchies (SPH) based on Ogma's whitepapers (AOgmaNeo)…
This report proposes a unified taxonomy and literature system for agentic computer graphics driven by large language models, analyzing trends and open challenges in human-agent collaboration and long-horizon planning.
Chris Alexiuk has a simple rule about open source AI: If you believe in it, be loud about it. “We should be shouting about this technology from the rooftops,” said Alexiuk, a Toronto-based senior product research engineer in deep learning at NVIDIA who…
Potential Sciences, a startup backed by the UK government, is betting that…
Yesterday's hot paper was Sparks of Weight-Space Programming: LLMs can natively read & write neural network weights by Rikiya Takehi, Jayden Teoh, Alex Zhang, and Omar Khattab. The paper (OK, the authors classify is as a "blog post", but it's paper-shaped)…
TurnWise improves streaming video understanding by assigning turn-level credit and balancing decision weights in group relative policy optimization.
Welcome to Import AI, a newsletter about AI research. Import AI runs on arXiv, cappuccinos, and feedback from readers. If you’d like to support this, please subscribe. Subscribe now LLMs are very bad at low-latency tasks: …Realtime-Bench says it’ll be a…
Small disclaimer first, this is one person on one codebase for one month, not a benchmark. I got annoyed at my usage bill so I tracked it for 4 weeks on our main app, a Next.js frontend and a Node API, roughly 60k…
Models for physical AI tasks, such as robotics and autonomous driving, need not reason solely in language. They can use visual plans--points, bounding boxes, and trajectories--to represent spatial goals and anticipate motion. However, autoregressive vision-language models…
医学论文写作与科研助手：面向中国医生和医学研究生的 32 个 Agent Skills，适用于 Claude Code、Codex，覆盖选题、PubMed 文献检索、中英写作、选刊投稿与审稿修回。Medical writing & research skills.
There’s a lot of dispute to yesterday’s title story on “ Weight Space Programming ” — and we agree that some of the claims are presented in overly favorable light and overall took away from the impact of the paper’s more substantial discoveries. However it…
Big update to the Delta Inference Engine, you guys asked for more features, more benchmarks and less clickbait. I'm here to deliver The short version: I've added almost everything requested and more. Delta Inference…
CoworkOS is a long-horizon agent harness that separates strategy from execution, supports human-in-the-loop collaboration via shared artifacts, and improves model performance through sustained automated post-training.
Tencent has unveiled Hy4, its first multimodal language model that understands images and can operate software on its own. On the Artificial Analysis Intelligence Index, it scores 35 points, solid but behind Kimi K3.1 Pro. On hard math and complex image…
NEO-VLA eliminates fixed visual encoders to let action policies access evolving visual representations across network depths, improving manipulation robustness.
I feel like LLMs have made it much easier to learn about almost any subject. That got me wondering: is anyone here using them every day to study a specific topic in depth, with the goal of becoming an expert in that…
Route Anything is a framework for dynamically selecting and coordinating specialized AI models and policies across diverse tasks to improve performance and efficiency.
While frontier AI models have solved a remarkable series of outstanding mathematics problems and made progress in the life sciences, they have yet to truly plumb the complex physics in advanced materials.
Amid a deepening bear market for AI/tech (worthy of dark humor ) with Anthropic bonds receiving a “junk” rating and OpenAI fundraise headlines , we continue to see very surprising capability finds in the community, such as today. The best way to put today’s…
Kling's multimodal understanding stack links caption density, perceptual text encoding, reward modeling, and evaluation protocols to improve controllable video generation.
MoviEdit is an end-to-end framework for long-form multi-shot video editing that uses cascaded bidirectional-autoregressive modeling and a new benchmark to maintain coherence across scenes while reducing manual data annotation.
67 more stories, by section and heat.
The hottest AI research stories from the previous 5 issues.