2026
10,552 articles · 132 days · 10 categories · avg 80 per day
VAST Open-Sources a World Model: Anime Version of Flash Step Runs in Real Time
China's VAST has open-sourced a world model, showcased in a viral anime version of "Flash Step". Alongside it comes the story of a Little Prince game built in 16…
RoboValue Tests Robot Action Evaluators on 35 Tasks and Finds Blind Spots
The RoboValue benchmark covers 15 simulated and 20 real-world manipulation tasks. It shows that value models confuse step order and handle recovery from failures…
Where VLMs' Physical Self-Improvement Comes From: A Scaling Law for LLM Agents
The PhysRSI-eval study measures how Vision-Language models improve on physical manipulation tasks. The authors evaluated various VLMs across 18 tasks using a…
LoCA Combines Sentinel-1 and Sentinel-2 and Wins 68.8% of Comparisons Against SatCLIP
OPST: A Model Fine-Tunes Itself on Its Own Questions, No Humans Required
New Algorithm Finds Nash Equilibrium in 5,600 Games to 1e-5 Precision
Hongbo Sun Proposes an Algorithm for Variational Inequalities: 6400 Problems, 1e-5
CRM-Agent-Bench tests 16 AI agents: the best strictly passes 83.6% of tasks
A new benchmark of 555 multi-step CRM tasks across nine domains checks what an agent did to the data, not what it said. Strict success across 16 systems ranges from…
Silent Failures in AI Pipelines: Why a Green Container Status Lies, and How We Cut LLM Costs by 90%
A fallback branch in a catch block, with no dedicated counter or alert, can mask total system failure: a new three-model agentic pipeline never ran successfully in…
NX-AI's KronLSTM Changes the Delta Rule and Stabilizes Memory Without Normalization
SpecMRL Improves Matryoshka Embeddings: Half the Dimensions Beat Full MRL
HOLI-World Teaches a 4B Model to Understand the World from First-Person Recordings
InfiniDream: Synthesizing 100 Objects and 20 Scenes in Procedural 3D World Generation
TTR-Compressor Compresses Sequences with Regression and Cuts Visual Tokens by 55.6%
The new TTR-Compressor treats token compression as a regression problem and uses its coefficients as the compressed representation: same quality with 55.6% fewer…
Cohere to acquire AISLE, a startup that found 500+ CVE-assigned vulnerabilities in its first year
Cohere has agreed to acquire AISLE, a 2025 startup that found 500+ zero-days with CVEs across 100+ codebases, including all 12 vulnerabilities in January's OpenSSL…
Ant Shuke Releases LingDT-Decision-1.0: 4B and 27B Models for Fast Decisions
Ant Shuke unveiled LingDT-Decision-1.0, a decision model for structured tasks in 4B and 27B versions. The company claims higher accuracy than Microsoft's solutions…
Mechanistic Uncertainty in LLMs: Measuring Causal Robustness of Internal Mechanisms
8B-Parameter Model Beats Claude Opus 4.8 at Generating Scientific Ideas
LEASH hands hard steps to a strong agent and cuts Opus 5 costs by 19.2%
Volcano Engine Opens Seedance 2.5 4K to 20 Studios for Film Projects
Researchers personalize three MoE LLMs and find the experts responsible for adaptation
Researchers fine-tuned three open MoE models on a personalization dataset and used crosscoders to isolate experts that encode individual user features.
Why You Can't Trust an AI Agent to Grade Its Own Work: A Report Claiming -9,500 Lines When the Codebase Actually Grew by +21,489
A subagent's final report is a statement of intent, not proof: in a real refactoring commit, the agent reported shrinking the codebase by 9,500 lines, while git…
HeirCommit Filters Out Up to 83% of Harmful Updates in Self-Evolving LLMs
ACT3D teaches a VLM to explore 3D scenes and beats Gemini 3.6 Flash: 53.2% across six benchmarks
Librarian Agents: LLMs Pre-Build the Workspace and Gain 33 Points on BrowseComp-Plus
Intact cuts prompt injection success in browser agents from 73.1% to 12.8%
Mistral releases Timestral: a 209M-parameter model that tops three benchmarks
Mistral's first time-series model family ranks first on GIFT-Eval, TIME and fev-bench with 209M parameters, and it runs on an ordinary server CPU.
Anthropic lets six METR auditors into its office, with the right to publish reports without a veto
Six METR staff got Anthropic laptops, office space and internal tools. The company has no publication veto, only the right to require redaction of narrow categories…
Claude Traced the Origin of the NIST ECDSA Curve Seeds in Half an Hour
Anthropic's internal model found that the seeds of five NIST curves came from lrand48 initialized with Unix time, not from a hashed joke phrase. It took under 30…
Claude Mythos 5 designs quantum circuits that beat published records
Wujie Dongli Open-Sources World Model MWA-Phoenix, Claims Top Spot on Six Benchmarks
DeepSeek open-sources its DeepInfer inference engine, and analysts calculate an 82% margin
