model releases

I audited 443 GGUF quants across 25 repos. 64 of them can't be the quant their filename claims.

TL;DR: k-quants need tensor rows divisible by 256. When they aren't, llama-quantize quietly swaps in a ~4.5 bpw type and the file keeps its low-bit name. I audited 443 quants across 25 repos; 64 are a

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

I benchmarked 9 open models on spotting fake sources during agentic search (DeepSeek V4, Qwen 3.8, Nemotron 3 Ultra)

DGX agent

I built a benchmark called EchoNet. An agent gets a factual question, then searches a syntheic web I made before answering. Some of that web is seeded with misinformation: one fake page, a fake page r

model-releasesr-localllama

28 Aug 2026

I fine-tuned Qwen3.5-4B on ~3k browser trajectories and improved benchmark results from 22% to 62%

Smaller models still struggle with browser tasks that require open-ended research or deep reasoning. But many practical browser-automation workflows are much more procedural: extracting structured dat

model-releasesr-ollama

I reverse-engineered an NPU vendor's engine format (int8 weights stored as two nibble planes) to run GGUFs with no model conversion — now 1.5× faster than the vendor's own runtime

I've been running Qwen3-0.6B on the M5Stack LLM-8850 card (Axera AX8850 NPU, 24 TOPS, 8GB LPDDR4x) hosted by a Raspberry Pi 5 — as a llama.cpp backend. The problem: the vendor stack requires convertin

I subscribed to Ollama Pro today. deepseek-v4-flash:0731 works great and serves as a solid alternative, and glm-5.3-flash:cloud works fine as well.

However, I keep getting a 404 error when trying to access qwen3.8-flash-next:125b-mlx and ornith-1.5:397b via the API, even though they are listed in the model list. Does anyone know why this is happe

model-releasesr-ollama

I used Hermes Agent + ComfyUI MCP + MiniMax prompts to turn a music track into a short music video

model-releasesr-stablediffusion

The initial reason for this was the Comfy H3 Sync & Sound Community Challenge: Comfy H3 Sync Sound Community Challenge! - by Allyson Toy I made a short rap track in Suno, then used Hermes Agent to bui

Invocation-Level Reliability of Tool-Using Agents

arXiv:2608.26189v1 Announce Type: new Abstract: Tool-using agents fail two ways: choosing the wrong tool, or forming wrong arguments, and an early failure of either kind can silently corrupt everythin

Is there a cache bug in the Claude Code CLI when using external models, or maybe even within the tool itself?

I read this somewhere and wasn't sure if it's still an issue or if it only affects external models versus all models used via the Claude Code CLI. 'This latest Ollama update finally fixes that issue w

model-releasesr-ollama

Just a rumour of a bug is enough to find a security exploit these days

model-releasessimon-willison

Just a rumour of a bug is enough to find a security exploit these days Anil Madhavapeddy is a professor of computer science at Cambridge and a core maintainer of the OCaml compiler. In this somewhat a

KinyaEmbed: Contrastive Sentence Embeddings for Kinyarwanda via Multi-Stage Curriculum Training

arXiv:2608.26941v1 Announce Type: new Abstract: We present KinyaEmbed, the first dedicated sentence embedding model for Kinyarwanda, a morphologically rich Bantu language spoken by over 12 million peo

KnockGS:interaction-Grounded Calibrationof Physical Gaussian Representations

arXiv:2608.27365v1 Announce Type: cross Abstract: Physics-integrated 3D Gaussian representations now allow reconstructed deformable objects to be simulated and rendered under explicit material models.

Knowledge Cards: Structured Knowledge for AI Systems

arXiv:2608.26176v1 Announce Type: new Abstract: AI systems whose outputs inform real decisions, and increasingly consequential ones, require something that current documentation practice does not prov

Knowledge-Verified Emergent Deception in LLM Agents Under Conflicting Incentives

arXiv:2608.26372v1 Announce Type: cross Abstract: Large language models are increasingly deployed as autonomous agents serving users on behalf of companies, placing them in settings where user and dep

Learning the ARTS of Search for Automated Discovery

arXiv:2606.21891v2 Announce Type: replace Abstract: Scientific discovery can be formulated as an iterative search process over the space of hypotheses and experiments. Contemporary methods navigate th

Legal Rule Induction: Towards Generalizable Principle Discovery from Analogous Judicial Precedents

arXiv:2505.14104v2 Announce Type: replace Abstract: Legal rules encompass not only codified statutes but also implicit adjudicatory principles derived from precedents that contain discretionary norms,

Leveraging Large Language Models for Systematic Literature Review of Disease Spread Models

arXiv:2608.26150v1 Announce Type: new Abstract: Recent advancements in Large Language Models (LLMs) have created new opportunities to streamline and potentially automate many research processes, inclu

Linear Independence of Polynomial Compositions and Identifiability of Deep Neural Networks

model-releasesarxiv-cs-lg

arXiv:2608.27113v1 Announce Type: cross Abstract: Motivated by theoretical problems in deep learning, we conjecture that post-composing a fixed number of pairwise distinct nonconstant polynomials with

LLaVAFlow: Preserving Latent Alignment Flow for Parameter-Efficient Multimodal Fine-Tuning

arXiv:2608.26820v1 Announce Type: new Abstract: While Multimodal Large Language Models (MLLMs) exhibit strong generalization, visual instruction tuning for downstream tasks inevitably causes catastrop

LLM Analysis of 150+ years of German Parliamentary Debates on Migration Reveals Shift from Post-War Solidarity to Anti-Solidarity in the Last Decade

arXiv:2509.07274v4 Announce Type: replace Abstract: Migration has been a core topic in German political debate, from postwar expellee displacement to labor migration and recent refugee movements. Larg

LLMs Can Design Near-Optimal OR Algorithms

arXiv:2608.27296v1 Announce Type: new Abstract: We ask whether large language models (LLMs) can design effective algorithms for well-specified operations research (OR) problems. We study inventory con

LLMs for Academic Workflows: An Evaluation of Literature Reviews Generated with Short and Long Context Windows of LLMs

arXiv:2608.26145v1 Announce Type: new Abstract: Our research focuses on evaluating literature reviews generated in short and long context settings of large language models (LLMs) to investigate the im

Local agentic coding Benchmark : Qwen3.8-Flash-Next NVFP4 vs 27B (and the others...)

Using https://huggingface.co/RadixArk/Qwen3.8-Flash-Next-NVFP4 and https://old.reddit.com/r/BlackwellPerformance/comments/1w04xb7/qwen38_flashnext_on_1x_rtx_pro_6000_171_ts_c1_428/ As usual, all the d

LoRA as Oracle

arXiv:2601.11207v2 Announce Type: replace-cross Abstract: Practitioners increasingly deploy neural networks they did not train, and must audit them after the fact for hidden backdoors, without the tra

LowRankArena: A Standardized Evaluation Platform for SVD-Based LLM Compression

arXiv:2608.26389v1 Announce Type: new Abstract: SVD-based low-rank compression has become a fast-growing direction for reducing the memory and computational cost of large language models (LLMs). Howev

MAPLE: Metadata Conditioned LLM Pretraining for Locale-Aware Question Answering

arXiv:2601.15236v2 Announce Type: replace Abstract: Large language models can memorize competing locale-specific facts yet fail to select among them when the locale changes, defaulting instead to a si

Mechanistic Reaction Prediction via Discrete Flow Matching on Graph-Structured Electron Occupation

arXiv:2608.27429v1 Announce Type: new Abstract: Chemical reactions are fundamentally transformations in electron space, yet most machine learning approaches model them either through extit{de novo} ge

Memory Anchors for Continual Robot Learning

model-releasesarxiv-cs-ro

arXiv:2608.26545v1 Announce Type: new Abstract: Robot policies deployed in the wild should have the capability to continually learn new tasks without forgetting existing behaviors. A common approach t

MemToC: Benchmarking Memory-Tool Conflict Resolution in Large Language Models

arXiv:2608.26295v1 Announce Type: cross Abstract: Tool-augmented LLMs must arbitrate between two fallible sources when a tool return conflicts with their parametric memory, yet existing evaluations me

Meta-Learning Where to Allocate Experts: Task-Conditioned Layer-Wise Compression for MoEs

arXiv:2608.26650v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models route each token to a subset of expert networks, increasing capacity while keeping per-token computation sparse. In many

Methodological and Conceptual Framework for 5D Multi-Table Analysis: A Unified Approach for Complex Data Reuse

arXiv:2608.26149v1 Announce Type: new Abstract: Multi-table learning remains a major challenge in machine learning for healthcare and other complex information systems. Relational data combine several

Modality Maturity Index: A benchmark for assessing multimodal capabilities of omni models

arXiv:2608.26317v1 Announce Type: cross Abstract: Frontier language models are increasingly marketed as omni systems that can perceive and respond across modalities. Existing evaluation frameworks, ho

Modular Expert Merging for Biomedical Retrieval

arXiv:2602.04731v2 Announce Type: replace Abstract: Adapting general-purpose LLMs into domain-specialized dense retrievers typically requires large-scale training on mixed-domain data. We show that me

MoganColBERT-TR: A Late-Interaction Multi-Vector Retrieval Model for Turkish

arXiv:2608.26344v1 Announce Type: new Abstract: We previously reported a ModernBERT encoder trained from scratch for Turkish (MoganBERT-TR) and a single-vector embedding model built on top of it (Moga

Multi-Dataset Inverse Problem Solving with Distributed Generative AI

model-releasesarxiv-cs-lg

arXiv:2608.26283v1 Announce Type: cross Abstract: Extracting a shared set of unknown, not directly measurable quantities from multiple, heterogeneous datasets is a common challenge across scientific d

Multi-Expert Conformal Risk Control for Pairwise LLM Judging in Open-Ended Dialogue

arXiv:2608.26529v1 Announce Type: new Abstract: In this paper, we explore multi-expert Conformal Risk Control (CRC) algorithms for pairwise LLM-as-a-Judge evaluation in open-ended dialogue. Our core i

Multi2AV-Safety: Benchmarking Safety in Multimodal-to-Audio-Video Generation

arXiv:2608.26535v1 Announce Type: new Abstract: Audio-video generation is rapidly moving from prompt-driven synthesis toward multimodal conditioning, where text, images, audio, and video can jointly s

MVC-Bench: Benchmarking Calibration of Medical Vision-Language Models

arXiv:2608.27004v1 Announce Type: new Abstract: Reliable evaluation of vision-language models (VLMs) and medical vision-language models (Medical-VLMs) requires calibrated confidence, particularly unde

NeuDonatello: Uncertainty-Aware Framework for Accurate Neural SDF Learning

arXiv:2608.26504v1 Announce Type: new Abstract: Neural surface reconstruction has emerged as a powerful paradigm for recovering high-quality 3D surfaces from multi-view images. However, recovering acc

Ninfer and a 5090 with 3.8 27B is making me cry tears of joy it's so good.

Built the latest and I'm getting as much as 220 tokens per second and averaging in the 170s, I can't get over it. If anyone on here is on that project, fuckkkin' chapeau man, really incredible job. I

Non-binary bottom-up constituency parsing without arity actions

arXiv:2607.10591v2 Announce Type: replace Abstract: Non-binary bottom-up constituency parsing commonly uses arity-specific reductions such as (extsc{Reduce-}X#k), which jointly specify the mother labe

Not Just Reason, Not Just Scan: Reinforcement Learning for Proactive Scientific Error Verification over Academic Paper

arXiv:2608.26596v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) are increasingly capable scientific assistants, yet they remain far from fully autonomous research. This transi

Omni-Interactive Universal Embedder

arXiv:2608.27044v1 Announce Type: new Abstract: Multimodal representation learning has been shifting from traditional two-tower architectures to large language model (LLM)-based embedders due to their

On Scope Classification and Current Knowledge-Editing Benchmarks: A Negative Result, with INLAY as a Gradient-Free Case Study

arXiv:2608.26292v1 Announce Type: cross Abstract: Every memory-based knowledge editor in the SERAC lineage depends on a scope decision: given a query, does a stored edit apply? We report that current

open source caught up because it's open

Proof is in the method honestly. Closed model labs need to constantly reinvent the wheel to keep lead. Open source has a bunch of independent labs practically working somewhat together. Eventually whe

Order Matters: A Chinese Multi-Panel Meme Benchmark for Vision-Language Reasoning

arXiv:2608.26866v1 Announce Type: new Abstract: Many multimodal tasks depend on how visual elements are ordered and composed, not only on recognizing them in isolation. Internet memes are a compact ca

Ornith-1.5-35B-A3B on 8 GB VRAM: I think I've found my sweet spot

A few days ago I posted asking what people considered the best local model for an 8 GB VRAM GPU. At the time, my personal sweet spot was Qwen3.6-35B-A3B, for agentic coding with Pi.dev. Well… Thanks t

OS-Marathon: Benchmarking Computer-Use Agents on Vast-Horizon, Repetitive Tasks

arXiv:2601.20650v3 Announce Type: replace Abstract: Vast-horizon, repetitive workflows are common in daily routines, e.g., processing expense reports from a stack of receipts, organising a collection

Our government is broken. We are broken. Whether we stay broke depends on whether enough of us reject and denounce this garbage AND those wh…

model-releasesanthropic--x

Patrick Skinner tweets that the U.S. government is “broken” and urges citizens to reject and denounce what he calls “garbage” and its supporters. He emphasizes collective action as necessary to avoid