Kimi K3 opens 2.8T weights, Galahad caches 180 problems, StateAct cuts cost 9×
Every URL the pipeline pulled into ranking for this issue — primary sources plus the supporting and contradicting findings each Researcher returned. Inline citations in the issue point back here.
Sources
Kimi K3: Open Frontier Intelligence huggingface.co
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 context window. Kimi K3 is built on Kimi Delta Attention and Attention Residuals, which improve information flow across sequence length and model depth. Together with Stable LatentMoE, which effectively activates 16 of 896 routed experts per token, and refined training and data recipes, these advances yield an approximately 2.5x improvement in o
A Frozen 12B Beats Frontier Models on Verified Work: 100% Accuracy, 0 Tokens, Bit-Exact, Forever huggingface.co
Improving a language model today means retraining it: enormous compute, a new opaque model each cycle, non-deterministic output. We take the opposite path: the model stays frozen, and a persistent memory of verified solutions grows beside it. Once a problem family is solved and has passed an independent verification step that never consults the answer key, every new instance of that family is answered at zero generation tokens, bit-exact, deterministically. Across 180 fresh instances spanning ni
StateAct: Program State, before Pixels, for Long-Horizon Computer-Use Agents huggingface.co
Computer-use agents are usually improved by strengthening perception: better models for reading a screenshot and choosing where to click. Yet a screenshot is only a lossy rendering of the underlying program state, e.g., the files, application backends, and DOM that hold the task data. Different states can produce the same pixels, while code can inspect and modify that state directly. StateAct is a code-first, multi-agent harness built around this distinction. Its main agent works directly with p
Characterizing Warp Divergence from Pascal to Blackwell huggingface.co
Cycle-accurate microbenchmarks across Ampere, Hopper, and both datacenter and consumer Blackwell chips test the assumption that Independent Thread Scheduling fixed warp divergence after Volta. Divergent paths still serialize linearly across every generation, with SASS analysis separating stable behavior from architectural shifts.
A controlled multi-turn environment dissects planning across pre-training, single-teacher, and multi-teacher on-policy distillation stages, replacing opaque internet data with precise supervision. The setup lets researchers trace how planning ability is acquired, shaped, and integrated in foundation model agents.
From Proprietary to Open-Source: Bridging the Distribution Gap via Multi-Agent Protocol Distillation in Agentic Search huggingface.co
Since proprietary teachers block logit-matching, the method distills reasoning trajectories through a multi-agent protocol that densifies supervision beyond sparse RL rewards. The approach narrows the distribution gap between closed and open models on knowledge-intensive agentic search, drawing 82 upvotes on Hugging Face.
Codifying the Judge: Scalable Evaluation via Program Distillation huggingface.co
Instead of calling an LLM at evaluation time, PAJAMA compiles judge decision logic into a committee of programs that score candidates directly. The programmatic judges cut cost and latency while making decisions transparent and editable, unlike opaque LLM-as-a-judge pipelines.
A Vocabulary for Multi-Agent Automated Research Systems huggingface.co
The framework specifies eight axes — agent roles, available operations, invocation rights, communication, information visibility, action selection, run initiation, and output evaluation — so trajectories from different automated research systems become directly comparable across papers and implementations.
Reasoning Denoiser: Denoising Reasoning Traces for Hallucination Detection in Large Reasoning Models huggingface.co
Long reasoning traces from large reasoning models hide truthfulness cues behind irrelevant and repetitive steps. The Reasoning Denoiser removes both noise types before hallucination detection, restoring signal that noisy trajectories otherwise degrade substantially.
Evidence Attribution in Visual Document Understanding without Coordinates or Region Labels huggingface.co
Coordinate-based evidence attribution triggers Attribution Hallucination, where vision-language models mark the wrong regions even when answers are correct. The method drops coordinates and region labels, letting models quote textual evidence and retrieve its location afterward for reliable visual document understanding.
TRACE: Business Rule-Grounded Reasoning Curriculum for Knowledge-Preserving Parametric Tool Retrieval in Enterprise LLMs huggingface.co
Parametric retrieval enables LLMs to retrieve tools implicitly by assigning each API a unique virtual token and training the model to generate it via constrained beam search. Toolsense shows that this regime has two critical drawbacks: it destroys parametric tool knowledge during training, and its beam-search decoding is too slow for real-time deployment. We introduce TRACE (Tool Retrieval via Augmented Chain-of-thought and Enterprise rules), a two-stage curriculum that resolves this dissociatio
WorldDiT: A Unified Diffusion Architecture for World and Action Modeling huggingface.co
Many recent robot policies pursue stronger control by using large pretrained vision-language models (VLMs) as the action backbone. We introduce WorldDiT, a unified diffusion transformer architecture that couples action generation with visual world modeling and achieves strong performance without a large pretrained VLM action backbone. During training, a single diffusion transformer generates continuous action chunks and predicts normalized RGB patch targets from future camera frames. Across four
Data Pyramid for Embodied Manipulation huggingface.co
Multimodal foundation models learned to see and to speak by consuming the whole internet. Embodied agents admit no such shortcut, since they require data that couple observations with physical states and actions. These signals can be provided, to varying degrees, by multiple data sources. In this work, we organize the embodied data ecosystem as a “pyramid” spanning five complementary sources: real-robot data, UMI-style data, egocentric and exocentric data, simulation data, and general vision-lan
DecoupleMix: Decoupled Ratio Search and Convex Allocation for Scalable VLM Data Recipes huggingface.co
While data curation for Vision Language Models (VLMs) is increasingly active, public practice for constructing pretraining mixtures remains largely heuristic: practitioners stack datasets that pass quality filters, set cross-domain ratios by intuition, and lack a principled, attributable criterion for admitting new data, while frontier recipes remain undisclosed. We formulate data construction as a systematic mixture-optimization problem and turn it into a reproducible engineering discipline by
Sol-Attn: Accelerating Video Generation Inference via On-the-Fly Attention Sparsification huggingface.co
Diffusion transformers are essential for high-fidelity video generation, but long token sequences make attention a dominant inference bottleneck. Training-free dynamic sparse attention alleviates this bottleneck by computing only selected key-value blocks, yet existing methods struggle to sparsify attention both efficiently and accurately for two reasons: (1) Rigid, unpredictable, and costly routing: selecting a fixed fraction of top-ranked blocks by proxy score imposes fixed budgets, whereas re
Progress Reward Modeling for Robotic Learning: A Comprehensive Survey huggingface.co
Robotic learning takes place in dynamic environments with large behavior spaces. A terminal success signal only tells the robot whether the task is completed. It does not explain whether the current behavior is making progress, remaining unchanged, or undoing earlier progress. For this reason, recent studies have increasingly explored progress rewards that provide feedback during task execution. However, the current literature lacks a shared framework. Existing methods use different observations
JarvisHub: An Open Harness for Canvas-Native Multimodal Creative Agents huggingface.co
Creative AI is moving from single-step asset generation toward long-horizon multimodal production. Although recent generative models can synthesize high-quality images, videos, audio clips, UI elements, storyboards, slides, and other creative assets, real-world creative work requires more than isolated prompt-output interactions. It involves references, drafts, alternatives, edits, failed attempts, version relations, tool actions, evaluation signals, and human feedback, which together form an ev
UltraViT: Latency-Optimized On-device Vision Encoder for Large Vision-Language Models huggingface.co
Large Vision-Language Models (LVLMs) remain bottlenecked by massive computational footprints, precluding their deployment on resource-constrained edge devices. While efforts to compress LVLMs focus heavily on vision token reduction or smaller language models, the vision encoder is largely overlooked, typically deployed as a monolithic, computationally heavy feature extractor. Moreover, there is no previous effort that designs a vision encoder for LVLMs directly optimized for on-device latency. I
Rethinking Classifier-Free Guidance in On-Policy Diffusion Distillation huggingface.co
On-policy distillation (OPD) adapts diffusion models by querying a teacher along trajectories generated by the current student, but how it should behave under classifier-free guidance (CFG), a default component of modern diffusion systems, remains poorly understood. Existing OPD methods naturally extend velocity matching to the CFG-composed prediction, directly matching teacher and student guided velocities. We show that this objective is under-identified at the branch level: positive- and negat
dRAE: Representation Autoencoder with Hyper-Spherical Codes huggingface.co
In this work, we aim to discretize the high-dimensional visual representations to bridge the gap with language models - a non-trivial challenge, as existing quantization methods suffer from codebook collapse, failing to scale while preserving semantic coherence. We identify the root cause as metric mismatch: standard Euclidean codebook objectives are fundamentally misaligned with the anisotropic geometry of representation space, leading to codebook embeddings with high-variance magnitude scales
OmniVAE: An Audio-Video VAE with Cross-Modal Alignment for Joint Generation huggingface.co
Recent generative models are moving beyond silent video or standalone audio synthesis toward the joint generation of synchronized audio and video. Despite this progress, jointly generating audio and video with fine-grained cross-modal correspondence remains challenging due to their fundamental structural differences. Most existing methods use audio and video VAEs trained separately. As a result, the two latent spaces lack cross-modal alignment, leaving the downstream generative model to learn cr
FilmBench: A Film-Grade Benchmark for Cinematic Video Generation huggingface.co
Progress in video generation keeps narrowing the visual gap between AI-generated and professionally produced footage, yet most benchmarks still draw prompts from web sources or LLM templates and score them with untrained, generic multimodal models. More fundamentally, their evaluation taxonomies remain rudimentary (overall visual quality, coarse text alignment and temporal smoothness) rather than the professional Cinematic Language criteria by which films are actually made and judged, so they as
TILT: Improving Compositional Generation in Diffusion Models with a Model-Intrinsic Reward huggingface.co
Recent advances in powerful text-to-image generation models have made it increasingly important to develop test-time methods that modify the sampling trajectory to produce images more faithful to complex compositional prompts. We present TILT, a training-free framework for compositional text-to-image generation via test-time reward alignment. We interpret compositional failures as overlap modes between joint and single-concept distributions, and define a reward that favors samples where all conc
ClinFusion: A Vision-Centric Multimodal LLM System for Holistic Medical Understanding huggingface.co
Multimodal large language models (MLLMs) hold immense potential to revolutionize clinical practice, yet deploying them in the medical domain is fundamentally a vision-centric challenge: models must absorb knowledge from heterogeneous 2D and 3D medical images, and evaluation protocols must align with radiologists’ clinical practice and provide an accurate, fine-grained and factualness-driven assessment. In this paper, we introduce ClinFusion, a vision-centric MLLM designed for holistic medical un
Leveraging External Knowledge for Historical Document Restoration via Retrieval-Augmented Large Language Models huggingface.co
Historical documents act as invaluable knowledge archives but often suffer from illegibility due to physical deterioration and damage. While existing restoration methods based on masked language modeling effectively utilize local context, they struggle to restore named entities that require external historical knowledge. To address this limitation, we introduce a novel framework for historical document restoration that leverages large language models with retrieval-augmented generation (RAG). By
Chamaileon: Cross-Context Binder Design with Contextualized Modeling and Mixed Sampling huggingface.co
The rapid evolution of generative models has unlocked new potentials in protein binder design, a pivotal task in structural biology, by facilitating end-to-end generation via joint sequence-structure modeling or hallucination. However, existing approaches are predominantly implemented under a single-target, single-state assumption, limiting their ability to model multi-target or multi-state interactions required for advanced function-oriented protein design. Here, we introduce Chamaileon, which
IndicTalk: A Large-Scale Persona-Based Multilingual Conversational Corpus for Indic Languages huggingface.co
Large Language Models (LLMs) have transformed conversational AI, yet high-quality multilingual code-mixed dialogue resources remain scarce, particularly for Indic languages where speakers naturally alternate between English and their native language in both native-script and Romanized forms. We present IndicTalk, one of the largest multilingual Indic code-mixed conversational corpora, comprising over 13,28,604 event-grounded multi-turn conversations across 18 language varieties covering 9 Indic
GNM Head: A Generative aNthropometric Model of the human head huggingface.co
Parametric models of the human head are essential tools traditionally used in computer vision and graphics for animation, rendering, and reconstruction. More recently, they serve as crucial conditioning signals within generative large vision models, allowing for tight spatial control of generated imagery. However, existing publicly available models are typically limited in anatomical scope, modeling only outer geometry while ignoring intra-oral and ocular structures, and frequently suffer from r
Oxygen-TryOn: Fashion-Native Foundation Model for Any-item Virtual Try-On huggingface.co
We present Oxygen-TryOn, a unified foundation model for any-item virtual try-on. Rather than repurposing a general-purpose image editor, Oxygen-TryOn is fashion-native, built for try-on through a dedicated data engine and try-on-specific training. Given one or more reference items (clean product shots or in-the-wild worn-on photos) and a single target subject image, it synthesizes a photorealistic image of the subject wearing the items across virtually any fashion category. Prior systems handle
DriveDNA: A Large-Scale Multimodal Naturalistic Driving Dataset and Benchmark for Driving Style Identification huggingface.co
Driving style captures stable, driver-specific patterns in how a vehicle is driven. In naturalistic data, however, this signal is hard to isolate because drivers are observed in different vehicles, on different roads, and under different conditions, so models may mistake vehicle- or situation-specific regularities for driver-specific style. We introduce DriveDNA, a large-scale naturalistic dataset and benchmark for personalized driving-style modeling, comprising 4,121 drives from 465 drivers acr
Bitcoin Price Direction Prediction via Regime-Aware Multi-Modal Fusion of Social Sentiment and Technical Features huggingface.co
Bitcoin price prediction on sub-daily timescales is a hard open problem in computational finance. Bitcoin exhibits fat-tailed returns, non-stationary dynamics, and a price discovery process influenced by social discourse on Reddit and Twitter. Conventional approaches fuse OHLCV technical features with sentiment via static concatenation, applying identical fusion weights regardless of market state. This is inconsistent with the behavioural finance literature, which shows that retail sentiment is
References
SCMP — Trump tech official accuses Moonshot of stealing from Anthropic scmp.com
White House science adviser Michael Kratsios accused Moonshot AI of ‘large-scale covert industrial distillation’ from Claude Fable 5, with Anthropic reporting roughly 24,000 fraudulent accounts generating over 16 million interactions used for distillation.
SCMP — Global AI experts push back on US distillation claims scmp.com
Researchers including Nathan Lambert note distillation is a standard industry technique; critics point to a ‘timeline problem’ — K3 shipped 15 days after Fable 5, an implausibly short window to distill from the newer model.
Yutori Scouts — UK AISI / CAISI joint cyber evaluation scouts.yutori.com
On ExploitBench, Kimi K3 scored 32.2%, well below the 76.2% average of leading US models, and failed to achieve arbitrary code execution on any of 41 Chrome V8 tasks — though it completed 17 of 32 steps in ‘The Last Ones’ network simulation and rarely refused offensive requests.
AI Weekly — Redis zero-day report aiweekly.co
A swarm of Kimi K3 agents reportedly identified 19 zero-days in Redis in 90 minutes and produced a working RCE against Redis 8.8.0 in 27 minutes; Redis confirmed seven underlying memory flaws and shipped emergency patches on 23 July 2026 but did not validate the autonomy or timing claims.
DigitalOcean — Self-hosting Kimi K3 digitalocean.com
The 1.4 TB MXFP4 checkpoint needs 12–16 H200/B200 GPUs just to load, with Moonshot recommending 64-accelerator ‘supernodes’; rented Blackwell serving is estimated at ~$38/hour ($27,800/month) and on-prem CAPEX runs $2.6M–$4.1M with >$130k/year in electricity.
Puter.com — Kimi K3 review (citing Zvi Mowshowitz) developer.puter.com
Zvi Mowshowitz calls K3 the strongest open model ‘purely in terms of raw capability’ but ‘jagged’ and likely several months behind closed frontier; MindStudio’s trap-designed engineering tests show K3 fails multi-step planning problems where Claude succeeds.
Rozanov & Rei, arXiv:2410.02810 (original ‘StateAct’, 2024) arxiv.org
StateAct enhances LLM agents through self-prompting and chain-of-states… improved by 10-14% on ALFWorld, 30% on Textcraft, and 7% on Webshop, establishing a new state-of-the-art for few-shot in-context learning on these benchmarks.
Snorkel AI — OSWorld 2.0 leaderboard snorkel.ai
Claude Opus 5 has established a new state-of-the-art on OSWorld 2.0 with a success rate of 70.6%… compared to Claude Opus 4.8 which peaked at 20.6%.
AI Weekly — ‘StateAct cuts computer-use agent cost 9x on OSWorld 2.0’ aiweekly.co
The main agent required the GUI subagent for only 1.1% of its total steps… narration-blind finish gate spawns a fresh agent that only sees the original task and machine access, verifying persisted results rather than trusting the agent’s account.
Inspired eLearning — ‘ForcedLeak Breaks Salesforce AI Agents’ inspiredelearning.com
The ForcedLeak vulnerability (CVSS 9.4) in Salesforce’s Agentforce allowed indirect prompt injections hidden in a Web-to-Lead form to trick agents into exfiltrating sensitive CRM data.
UsageBox — ‘Cost per task: workhorse models 2026’ usagebox.com
A standard 1080p screenshot consumes roughly 16,000 tokens, while a 4K display can exceed 50,000 tokens per action… developers use PreToolUse hooks to trim redundant file reads, cutting total consumption by up to 90%.
Medium (Adnan Masood) — ‘The Hardest Easy Problem in AI: Computer-Use Agents’ medium.com
Agents no longer struggle with basic GUI controls or syntax; they frequently fail by losing track of long-term constraints, skipping necessary verifications, or failing to recover hidden states during multi-step processes.
AI Weekly — ‘Gemma-4-12B tops 31B on AIME via KV cache grafting trick’ aiweekly.co
grafting ‘verified solution libraries’ allows the 12B model to answer in as few as 61 tokens instead of re-processing a 400,000+ token reasoning chain, cited as up to 8,700× energy efficiency gain
ArxivLens popular-papers tracker arxivlens.com
flags the project with two ‘risk flags,’ citing ‘sparse artifact coverage’ and a ‘low confidence’ utility signal … estimates a multi-day reproduction timeline for meaningful verification of the results
TinyCrafts Chronicle (2026-07-19 daily) chronicle.tinycrafts.ai
claims are ‘big enough to deserve a careful read,’ but they currently stem from a single author and self-reported benchmarks; approximate similarity retrieval reportedly fails 94.3% of the time on larger stores
ResearchGate — Schelpe, ‘Byte-Exact Deduplication in RAG’ (Merlin paper) researchgate.net
SIMD-friendly open-addressing flat hash set paired with xxHash3-64 … sustained 8.7 GB/s, median in-process latency ~1.1 µs; input-size reductions from 13.9% (low-redundancy) to >71% (WildChat-like)
arXiv 2607.14431 — Schelpe, KV-cache grafting prior work arxiv.org
‘own-position graft’ is identified as the unique numerically exact operating point; shifting a KV block to a different absolute position introduces RoPE floating-point residuals that break bit-exactness
AIModeling.com news writeup aimodeling.com
byte-exactness verified only under a ‘pinned deterministic configuration’ on NVIDIA Blackwell and Hopper — moving a byte-identical store between machines is only guaranteed if they share the same underlying architecture
