Search: “agents”

Wild paper from Microsoft and colleagues. They show a new attack that reconstructs the text a local LLM generates by watching CPU cache activity while it detokenizes. Earlier cache attacks needed some

Found this post:https://www.reddit.com/r/ClaudeAI/s/At9QNEbawK of someone fully developing a game through AI writing code, generating 3D models, etc. (he is using Meshy to create the 3D models). I was

Bravo to Mistral! So great to see them killing it, they are an inspiration to everyone as we need more sovereign & open-source AI everywhere around the world. Excited to keep supporting them on https:

I am pleased to see that OpenAI’s new model is roughly on par with Gemini Flash and Opus 4.8 on prompt injection risk. Nice work! Evaluating and naming other labs turns out to be a great way to encour

Ling-3.0-flash-VL inherits the language, reasoning, and long-context capabilities of Ling-3.0-flash, while extending them with native image and video understanding. The model has 124B total parameters

I was reading up on the recent controversy around Tristan Buckmaster, Levent Alpöge, OpenAI, and the Navier–Stokes result, and it got me thinking about something broader than this particular dispute.

We have just raised 3B€ to accelerate our R&D. We are scaling up our training and inference compute to build future generations of our models, and expanding our Science team across all our offices --

I'm looking for more information to expand my creativity around this. I don't really have a realistic idea of what people actually do with local AI yet, I mostly just want to explore the possibilities

arXiv:2609.05258v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used to formulate optimization models from natural-language problem descriptions, yet realistic operatio

arXiv:2609.04206v1 Announce Type: cross Abstract: Voice AI systems increasingly mediate customer care interactions where caller presentation cues such as accent, affect, fluency, and urgency are avail

Fascinating paper from Anthropic and colleagues. They study whether models can tell when they are being tested. It turns out that capable models can tell when they are being tested rather than deploye

I used an LLM to iteratively evolve an optimization algorithm rather than solve the packing directly. Starting from a simple seed solver, the LLM proposes algorithmic changes guided by a scoreboard of

The Guardian: Matt Clifford steps down as the chair of the UK government's science and tech research unit after joining Anthropic, following conflict of interest concerns — Matt Clifford forced to sta

DISCLAIMER I ramble a lot and am not good at focusing on documentation and or coherent thoughts. TLDR I did things I made a thing and it works for me. I will try to answer any questions. A few people

How it works. The model being tested is given a server capable of running it's weights and full context. That server is placed in a median priced apartment. The AI is given a bank account with for ren

While we see for frontier models similar scores among famous coding benchmarks, across: DeepSWE, Terminal-Bench, LiveCodeBench, Code-Arena ELO. Here are in my opinion some next level benchmarks that r

Hi, I've been wanting to run my own local LLM for some time and I finally saved enough to get a budget GPU. I can spend 700 at most, and I'm looking for a GPU that can run quantized 30B~ parameter mod

Today we're releasing data on models accelerating research at OpenAI. Recursive self-improvement could be the most important contributor to AI capabilities over the next few years, but by default it w

Anyone who has a 3D printer and get use of it finds it incredibly useful for those odd jobs around the house, a missing bracket, a cable router, steam deck holder and so on. In the past if I was missi

me: AI isn’t that smart & remains unreliable and now I am worried that monitoring-which likely could have prevented multiple incidents-is being reduced, and also worried that OpenAI has consistently n

I've been running Qwen3.8-27B as a local inference server for a production content intelligence pipeline (HVAC industry stuff, lots of long-context retrieval and structured extraction). I have been wa

100%, which is the core of why I called for a pause on OpenAI in my newsletter today. The Hugging Face incident was probably just the tip of the iceberg. OpenAI has lost control and they're hiding imp

arXiv:2609.04177v1 Announce Type: new Abstract: Explaining why a specific outcome occurred, and which inputs deserve the blame or credit, is central to philosophical, scientific, and policy analysis.

Chase DiFeliciantonio / Politico: California AG Rob Bonta is investigating OpenAI over the Hugging Face hack in July, after more than a dozen states joined Alabama in its investigation — SAN FRANCISCO

arXiv:2609.03721v1 Announce Type: cross Abstract: Context: Architectural Design Decisions (ADDs) capture the rationale behind the structure and evolution of software systems but are rarely documented

Despite everything that has happened, OpenAI has seemingly decided to stay competitive by burning down the only meaningful bit of safety assurance we actually have today - CoT monitoring. Completely d

arXiv:2609.03611v1 Announce Type: cross Abstract: Vision-Language Models (VLMs) are increasingly used to evaluate robot manipulation outcomes, but existing benchmarks offer limited evidence of cross-d

arXiv:2609.02984v1 Announce Type: new Abstract: The conventional approach to machine learning, that is, collecting data, training models, and performing inference in a single location, faces fundament

Gemini 3.8 Flash might be maxxed on other benchmarks, but not quite on document parsing. We benchmarked it on ParseBench. It does slightly better on tables + content faithfulness compared to Gemini 3.

We all want more tokens per second but I keep seeing confusion on what to expect for given hardware. For the decoding phase (TG/s) to produce one token all the model weights and KV cache needs to be r

After Qwen3.8 27B came out, I decided to benchmark the models that could fit in my GPU (RTX 5080) on my actual code (C code), the results were not completely unexpected but some quants were definitely

arXiv:2609.03796v1 Announce Type: cross Abstract: We introduce LLaDA-Image, a unified framework that pairs a 6B Diffusion Transformer (DiT) trained from scratch with a frozen vision-language understan

arXiv:2609.03660v1 Announce Type: cross Abstract: The dominance of Neural Networks (NNs) in RL is partially due to their incremental learning capability, which naturally suits the online, non-stationa

arXiv:2609.03507v1 Announce Type: cross Abstract: Tracking depression from multi-session counseling dialogues requires estimating both current symptom severity and how it changes across sessions. Yet

arXiv:2609.03209v1 Announce Type: new Abstract: We study a governed approach to enterprise analytics: a language model interprets the question, while deterministic policy selects and runs a pre-approv

New from the LangSmith Signal. Over the last 2 weeks, we looked at which models teams reach for, and which ones are doing the work. ✅ Reach: gpt-4o-mini was used by 13% of orgs ✅ Work: gpt-4.1-mini sa

🦙 @ollama is used by 9 million developers and 85% of the Fortune 500, giving co-founder and CEO Jeffrey Morgan (@jmorgan) a unique view into which AI models people are actually using and how that’s ch

arXiv:2609.03293v1 Announce Type: new Abstract: Personalized assistants should not only comply with user requests but also assess whether those requests are appropriate given the user's current circum

arXiv:2609.03487v1 Announce Type: cross Abstract: Knowledge graph embedding (KGE) demonstrates its effectiveness for predicting missing links in knowledge graphs (KGs) by projecting entities and relat

arXiv:2609.04151v1 Announce Type: new Abstract: Generative image models can now produce high-quality images, follow complex instructions, and support precise edits, but they still struggle to preserve

arXiv:2609.03702v1 Announce Type: new Abstract: General-purpose code embeddings power tools for code search, classification, and retrieval. Compact transformer encoders for code typically rely on eith

Unlike other players in the OCR space, we don't block our competitors from accessing our platform. We're also fully committed towards open, reproducible benchmarks for document parsing and extraction

I was using: weights https://huggingface.co/RadixArk/Qwen3.8-Flash-Next-NVFP4 with the optimized SGLANG (patched) from https://old.reddit.com/r/BlackwellPerformance/comments/1w04xb7/qwen38_flashnext_o

Overview llama.cpp 0.4.0 adds initial Qwen3.8-Flash-Next and Nemotron-3-Puzzle support, on-demand tensor reading, per-slot server context limits, video input options, and a ggml update to 0.23.0 with

arXiv:2609.03949v1 Announce Type: cross Abstract: The problem. A long-lived KV cache must be compressed before the queries that will read it exist; selection by observed attention (H2O, SnapKV) collap

arXiv:2604.07119v2 Announce Type: replace Abstract: Peer review plays a central role in the NLP publication process, but is susceptible to various biases. Here, we study language-of-study (LoS) bias: