model releases

misc : prevent RAM peaking at model loading stage (#27483) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44939788 macOS/iOS: macOS Apple Silicon (arm64) m

sycl : enhance the api to support peer-to-peer copy (#27550) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44948405 macOS/iOS: macOS Apple Silicon (arm64)

vulkan: fix FA dequant path engagement (#28190) Skip the nb[3] check when ne[3] == 1, the shader never reads it for a single stream. Cache views carry the full-buffer stride there, so the old check re

CUDA: Allow concurrent streams per split for multi-GPU (#28198) CUDA: Allow CUDA optimization per split for multi-GPU. Previous guard caused multi-GPU to skip the graph optimization. The graph is alre

mtmd : add const in various places (#28307) mtmd : mark context as const in more methods Mark mtmd_context as const in: mtmd_bitmap_init_lazy mtmd_tokenize mtmd_tokenize_from_parts mtmd_helper_support

metal : fix glu dispatch with ne00 = 1 (#28306) metal : fix glu dispatch with ne00 = 1 tests : disable ill-defined tests Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/

metal : add sparse FA (#28098) metal : support n_kv_max sparse mask hint in flash attention vec kernel add kernel_flash_attn_ext_vec_idx: compacts finite mask entries into a per-row index list (Hillis

mtmd: propagate const to preproc class (#28310) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/44990064 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple

tune MMVQ to MMQ crossover for SM87 (#28285) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/45087167 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Sil

opencl: quant lm_head / decode GEMV and medium-batch GEMM optimizations (speculative decoding/MTP) (#26477) opencl: quant lm_head / decode GEMV and medium-batch GEMM optimizations opencl: guard q4_K/q

common/json-schema : fix GBNF grammar generation for empty object schemas (#28279) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/45102223 macOS/iOS: macOS

llama: fix whole source code rebuilt on each new commit (#28278) Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/45114012 macOS/iOS: macOS Apple Silicon (ar

Banger paper from Google DeepMind and colleagues. (bookmark it) A model reads its entire KV cache on every generated token, even though it ends up attending to a tiny slice of it. In other words, if y

arXiv:2609.02512v1 Announce Type: new Abstract: Beauty assessments from Multimodal Large Language Models (MLLMs) are increasingly popular amongst users, companies, and aestheticians. This raises the q

arXiv:2609.01861v1 Announce Type: new Abstract: The performance of an LLM agent depends on the scaffold around a frozen model. A common way to improve that scaffold is to use a coding agent as an opti

arXiv:2609.01982v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as assistants for statistical and data science work, yet existing evaluations largely assume the anal

arXiv:2609.02831v1 Announce Type: new Abstract: Modern cameras transform RAW sensor measurements into sRGB images through an image signal processor (ISP). We benchmark two placements for blind restora

arXiv:2605.25784v2 Announce Type: replace Abstract: Multimodal large language models (MLLMs) have advanced rapidly in remote-sensing analysis, yet existing evaluations remain predominantly 2D-centric.

arXiv:2609.02502v1 Announce Type: cross Abstract: Text-to-image (T2I) models have achieved remarkable success at faithfully rendering specified objects and attributes, yet their ability to produce vis

arXiv:2603.06698v3 Announce Type: replace Abstract: Knowledge distillation between asymmetric architectures often induces severe geometric constraints on the learned representation space. We investiga

arXiv:2602.22453v3 Announce Type: replace Abstract: Recent work has identified a subset of attention heads in Transformer as retrieval heads, which are responsible for retrieving information from the

arXiv:2609.02131v1 Announce Type: cross Abstract: Sentiment in social-media threads does not only vary across posts; it shifts as users react to claims, corrections, evidence, and hostility within a b

Can a strong general-purpose vision model distinguish real from AI-generated images using only a simple decision rule on frozen representations? Our latest results, to be presented at #ECCV2026, show

Agentic artificial intelligence security startup Capsule Security Ltd. today released a detection system built on two Nvidia Corp. Nemotron models it fine-tuned itself, in what it calls an “AI circuit

arXiv:2609.02265v1 Announce Type: new Abstract: Personalized language agents use persistent memory to adapt to users over time, but the same mechanism creates an attack surface. When new information c

arXiv:2607.10110v2 Announce Type: replace Abstract: Recent work on looped language models suggests that many reasoning problems benefit from greater computational depth rather than from additional ind

OpenAI's ChatGPT, xAI's Grok, and Anthropic's Claude are back online after they all began experiencing issues around the same time on Thursday. At about 11AM ET, ChatGPT started returning error messag

arXiv:2609.02074v1 Announce Type: new Abstract: Planning is a central capability that enables agents to decompose complex long-horizon tasks into manageable steps. Test-time search and training-based

arXiv:2609.02735v1 Announce Type: new Abstract: Per-patient adapters are the preferred production architecture for dysarthric automatic speech recognition (ASR), yet parameter-efficient fine-tuning (P

arXiv:2609.01846v1 Announce Type: new Abstract: Recorded lecture videos, often enhanced with search and summarization features, are a standard study resource. However, students cannot easily ask cours

arXiv:2609.02459v1 Announce Type: new Abstract: We present CivBench, an open-source benchmark for evaluating language model agents in long-horizon, tool-mediated environments through the Model Context

Cloudflare Inc. today opened early access to Vulnerability Discovery and Remediation, a service that uses OpenAI Group PBC’s cybersecurity models to find software flaws in customer applications and bl

arXiv:2609.02273v1 Announce Type: new Abstract: Model merging provides an efficient paradigm for constructing multi-task large language models (LLMs) without full model retraining, yet it remains chal

Two weeks, one rule, and hundreds of entries from nearly 50 countries. Here's who took the four titles, and a look at everyone who made the final ten! Entries opened August 20 and closed September 1.

arXiv:2512.07328v2 Announce Type: replace-cross Abstract: Text-to-video generation has advanced rapidly, yet preserving a character's holistic appearance from a single reference image remains challeng

arXiv:2609.01952v1 Announce Type: cross Abstract: Decentralized, serverless learning increasingly connects devices running different architectures, where the standard tool, decentralized SGD, is undef

arXiv:2609.02730v1 Announce Type: new Abstract: Production recommender systems shape what billions of people see, and sustaining their performance requires continual optimization: as content, user beh

arXiv:2609.02316v1 Announce Type: cross Abstract: Generative engine optimization (GEO) enables content producers to increase the visibility of their web pages in generative search engines, but the sam

arXiv:2609.01787v1 Announce Type: new Abstract: Vision Transformers (ViT) excel in semantic understanding but fail to discriminate between object instances (e.g., identical embeddings for two dogs), l

arXiv:2609.01802v1 Announce Type: new Abstract: Prompt tuning provides a parameter-efficient way to adapt foundation models (FMs) by freezing the pretrained backbone and updating only a small set of l

arXiv:2609.01959v1 Announce Type: cross Abstract: Multi-contrast MR scans contain redundant structural information that can be leveraged during reconstruction and potentially accelerate acquisition ti

arXiv:2609.02573v1 Announce Type: new Abstract: Current evaluations and training of multimodal models predominantly focus on multi-image tasks, largely overlooking interleaved text-image scenarios. In

I've tried both and been having this debate with myself for the last few days, on two Asus Ascent GX10s (effectively the same as 2x DGX Spark): DeepSeek-V4-Flash-0731 (official weights) GLM-5.3-Flash

arXiv:2609.01795v1 Announce Type: new Abstract: Vision-language object detectors (VLODs) achieve strong zero-shot performance but remain vulnerable to distribution shifts during deployment. Mean-teach

arXiv:2609.01775v1 Announce Type: cross Abstract: Automated repair of Hardware Description Language (HDL) designs remains challenging due to the large search space of candidate repairs and the strict

arXiv:2609.01609v1 Announce Type: new Abstract: While diffusion models effectively capture multimodal behavioral priors for autonomous driving, offline reinforcement learning (RL) policies remain susc