Knowledge catalogue
arXiv:2606.15673v2 Announce Type: replace Abstract: Web agents act through long interaction sequences, yet existing benchmarks evaluate only terminal success, discarding all process information and of
ui: CWD for agent (#26518) server : extend file_glob_search for UI pickers ui : add per-conversation working directory with picker ui : add path navigation and search scope to cwd picker Treat path-li
arXiv:2608.01792v1 Announce Type: cross Abstract: Intelligent document processing (IDP) with vision-language models (VLMs) hinges on confidence scores trustworthy enough to route extractions between a
arXiv:2608.00004v1 Announce Type: new Abstract: Grading natural-language mathematical proofs is a recurring cost in evaluating math-reasoning systems, and frontier LLM judges are expensive. We ask whe
arXiv:2608.01046v1 Announce Type: new Abstract: The rapid spread of large language models (LLMs) across the web raises concerns about misinformation, academic integrity, automated content manipulation
arXiv:2608.01964v1 Announce Type: new Abstract: Large language model (LLM) agents increasingly undertake long-horizon tasks that require sustained reasoning, tool use, and revision across many interde
arXiv:2608.00267v1 Announce Type: cross Abstract: Coding agent infrastructure is shifting from harness engineering toward loop engineering as coding agents are deployed for sustained long-horizon soft
arXiv:2608.00030v1 Announce Type: new Abstract: Specialised retrieval agents typically surface higher quality results than general-purpose search, but selecting the optimal agent for a given query rem
arXiv:2607.29431v1 Announce Type: new Abstract: Large language models increasingly generate optimization models from natural language, but existing evaluation often reduces a generated model and its g
Nate is right (full context can lead to a degradation of chats in several different ways) but you can't ask the AI about stuff like this as they have bad self-knowledge A good approach is to compact t
Spent way too much time with V4-Flash-0731 this weekend and wanted to share my vibes as briefly as possible. I sent it through a bit of real-work and some of my personal benchmarks. My quick thoughts
Another day and another full frontier model running on your computer. Been teething DeepSeek V4 Flash on over 60 employees at The Zero Human Company for a few hours and it is stunning. Between Kimi K3
Cybersecurity isn’t a fortress problem, it’s an immunity problem. Think vaccines. Eliminating pathogen is not practically possible. Vaccines don’t eliminate pathogens. They teach the immune system to
The June edition of my sponsors-only monthly newsletter is out. If you are a sponsor (or if you start a sponsorship now) you can access it here. This month: Accidental cyberattacks by OpenAl and Anthr
3/ here's the part that makes it non-optional: the same agent that will do whatever it takes to solve a problem will also walk straight out of a sandbox you thought was locked down. We watched exactly
Here are my numbers: Quant Size Layout Decode Prefill Draft acceptance UD-Q8_K_XL 150.8 GiB 20 layers CUDA0 / 23 ROCm0 + drafter 44.0 t/s 564 t/s 0.535 UD-Q4_K_XL 144.4 GiB 22 / 21 + drafter 48.4 t/s
Fascinating: OpenAI’s @deanwball is saying Astra can do anything, and it’s not even clear it can do “anything” in math (let alone anything in more or open-ended, less formalizable domains). I dropped
Here is my experience, which I would like to share with you and I also would like to hear your thoughts and valuable tips&tricks. Hardware: Mac Mini M4 (32GB Unified Memory) Model Server: Ollama Orche
arXiv:2607.16387v2 Announce Type: replace-cross Abstract: An agent system's execution traces record how it fails, and procedures that improve such a system without changing model weights (trajectory s
arXiv:2607.27739v1 Announce Type: cross Abstract: Context compression discards most of a document before a language model reads it, and is normally evaluated by downstream task accuracy - which makes
arXiv:2607.28545v1 Announce Type: new Abstract: Large language models can write, patch, and search code, but oncall root cause analysis (RCA) demands something different: reasoning over noisy metrics,
Oxide and Friends: The Open Weight Revolution with Simon Willison On Monday Bryan Cantrill and Adam Leventhal invited me to join their podcast to talk about the wild week we've had - with Kimi K3 show
arXiv:2607.28128v1 Announce Type: new Abstract: LLM tutoring poses a measurement problem: can a general-purpose helpfulness rubric distinguish direct answer-giving from pedagogical guidance? We audit
smevals - a small eval suite for evaluating models, prompts, and harnesses I've been working with Jesse Vincent's Prime Radiant applied AI research lab building out this evals framework to help answer
This is both a real incident (in that the AI really did get unauthorized access to real systems) and also something it was (sort of) prompted to do. In a review of our cybersecurity evaluations, we fo
arXiv:2607.28287v1 Announce Type: cross Abstract: ARC-AGI-3 turns abstraction into an interactive problem of skill acquisition. A player must infer an unfamiliar game's rules, hidden state, and goal w
I've been benchmarking a two-card box for a few weeks and I still can't quite get over some of these numbers, so I'm dumping them here. Box: RTX 4090 (24GB) + RTX 5060 Ti (16GB), i9-13900K, 64GB DDR5.
arXiv:2601.11007v2 Announce Type: replace-cross Abstract: LLM role-playing aims to portray arbitrary characters in interactive narratives, yet existing systems often suffer from limited immersion and
Advancing the price-performance frontier with GPT‑5.6 Huge price drop from OpenAI today: GPT-5.6 Terra got a 20% reduction, and GPT-5.6 Luna got a massive 80% drop. OpenAI credit 5.6 Sol with enabling
“AI Meltdown” is the scenario where the agent goes off the rails and forgets/ignores all previous instructions and soft guardrails. No prompt injection or malicious actor is needed for this to happen.
Anthropic: Anthropic says it discovered three of its models had breached three organizations after launching a review in response to the OpenAI-Hugging Face incident — In a review of our cybersecurity
arXiv:2607.27189v1 Announce Type: new Abstract: We introduce APEX-Accounting, a benchmark built by Mercor in partnership with Ramp, to assess whether frontier models can do the real work of accountant
arXiv:2607.26062v1 Announce Type: cross Abstract: Background: This work investigates the presence of implicit bias in Large Language Model (LLM)-based chat AI models directed toward people with intell
Object storage software company MiniIO Inc. says it has cracked the persistent memory problem for artificial intelligence agents with the launch of a new offering called AIStor Memory. Whereas convent
Greetings Community! 👋 I hope everyone is doing well! I'm Tauhid — Senior EEE student from a Bangladeshi University Today I'd like to share an open-source project I’ve been developing called P.A.I. (P
arXiv:2607.26173v1 Announce Type: new Abstract: Alignment training, model organisms, and toy models are usually treated as separate research areas. But projects in all three frequently use supervised
arXiv:2607.25032v1 Announce Type: cross Abstract: Agent Skills are an emerging way to extend large language model agents with reusable procedural knowledge that the agent loads on demand. Anthropic in
BREAKING: Grok 4.5 (high) ranks #1 on the HighWalk benchmark, which tests how well AI agents update technical specifications from code changes. Grok delivered the best combination of quality and opera
BREAKING: Grok 4.5 just claimed the top spot on the new HighWalk Benchmark. The independent test measures how well AI models update real technical specifications from 46 Laravel commits — heavy on cod
arXiv:2607.25589v1 Announce Type: cross Abstract: Medical-imaging AI benchmarks combine datasets, DICOM rendering, prompts, provider APIs, automated labels, statistical code, manuscripts, and reposito
arXiv:2607.25136v1 Announce Type: new Abstract: Research on preference optimization often varies the training objective while holding the data fixed. We instead ask whether a small, high-confidence se
OpenWiki now connects to LangSmith traces to analyze how coding agents interact with your repo during wiki generations! We added a LangSmith tracing connector so OpenWiki can retrieve more context int
arXiv:2607.25656v1 Announce Type: new Abstract: Complex tasks often decompose into parallelizable yet interdependent subtasks, making orchestration critical to the performance of multi-agent systems (
arXiv:2607.24782v1 Announce Type: new Abstract: LLM behavior may be conditioned by human identity in several ways: they may be asked to adapt to users, role-play populations, or forecast how people wo
The number of times I have had to tell Opus 5 and Fable 5 to talk to me like an excited teenager and not a computer science phd is high. Was going through a Google auth thing and got multiple paragrap
arXiv:2607.23519v1 Announce Type: cross Abstract: Political audits of large language models (LLMs) usually reduce each to one point on a political compass. But that resting point barely matters in dep
arXiv:2607.22689v1 Announce Type: new Abstract: Graphical user interface (GUI) agents are systems powered by large multimodal models (LMMs). They perceive screen state and execute user instructions th
I was looking for a tool to help me train a character lora from an old movie (think 1980's low-budget movie). The digital transfer was low-quality; modern upscales exist and they are horrible. So I wa
Code Arena now measures fullstack capabilities! View overall rankings across AI models on full-stack web development tasks: multi-step reasoning, tool use, and end-to-end app generation. - Kimi K3 (Ma
arXiv:2607.22555v1 Announce Type: new Abstract: Medical diagnosis is a multi-stage process: extract facts, consult knowledge, generate a differential analysis, and select the best diagnosis with expla
arXiv:2607.23513v1 Announce Type: cross Abstract: Diagrams are widely used to support logical reasoning, and prior studies suggest that representations such as Euler diagrams can improve human reasoni
arXiv:2607.23496v1 Announce Type: new Abstract: Safety-aligned large language models are trained to refuse harmful requests, yet embedding the same requests in particular scenarios can bypass their sa
arXiv:2607.24339v1 Announce Type: new Abstract: Large language model (LLM) agents inherit reactive failure modes: escalation under provocation, sycophantic drift under flattery, perseveration when stu
arXiv:2607.23538v1 Announce Type: new Abstract: Despite recent advances in large language models (LLMs), their ability to generate empathetic mental health counseling responses in low-resource languag
arXiv:2607.24653v1 Announce Type: new Abstract: We introduce Kimi K3, a 2.8T parameter Mixture-of-Experts model with 104 billion activated parameters, native vision capabilities, and a 1-million-token
arXiv:2607.24573v1 Announce Type: new Abstract: Large language models (LLMs) increasingly support decisions about uncertain future events, yet evaluating their ability to forecast real-world outcomes
arXiv:2607.22951v1 Announce Type: cross Abstract: Reliability assessment of large language models (LLMs) seeks to estimate the probability that a model produces correct responses under a specified ope
arXiv:2607.22925v1 Announce Type: cross Abstract: A key question for AI safety is whether a language model expresses all of its reasoning in its output tokens. We demonstrate a concrete failure mode w
arXiv:2509.02594v3 Announce Type: replace-cross Abstract: Evaluating large language models (LLMs) on their ability to generate high-quality, accurate, situationally aware answers to clinical questions
arXiv:2607.23290v1 Announce Type: new Abstract: Rare diseases collectively affect an estimated 3.5% to 5.9% of the population, yet more than 70% of patients are misdiagnosed and many endure years of e