AI Research & Analytics
-
PhysX-Omni generates simulation-ready 3D objects — rigid, deformable, and articulated — for embodied AI
PhysX-Omni is a unified framework for generating simulation-ready physical 3D assets across object types: rigid bodies, deformable objects, and articulated objects (things with joints, like doors or robot arms). “Simulation-ready” is the key phrase — the outputs aren’t just pretty meshes, they carry the physical properties a simulator needs. ## What’s under the hood A… Continue reading
-
ACC compiles AI agent trajectories into long-context training data — turning agent runs into a learning signal
ACC (Agent Context Compilation) tackles a clean idea: when agents solve problems, they produce massive trajectories — tool calls, environment observations, dozens of turns. That data is usually thrown away after the task. ACC converts it into long-context training material instead. ## How it works When an agent works a search, software-engineering, or database-querying task,… Continue reading
-
FinceptTerminal is an open-source Bloomberg alternative with built-in AI agents and Python quant research
FinceptTerminal is an open-source financial terminal — a free alternative to Bloomberg that bundles market data, quant research, trading workflows, and AI agents into one desktop platform. Trending on GitHub at 537 stars per day. ## What’s in it Version 4 is built with C++20 and Qt6 as a native desktop app, embedding the full… Continue reading
-
Anthropic’s Project Glasswing has found 10,000+ critical vulnerabilities, partners report 10x bug-discovery gains
Anthropic shared a sweeping update on Project Glasswing — its AI-assisted security testing initiative powered by Claude Mythos. The headline number: more than 10,000 high- or critical-severity vulnerabilities uncovered across widely used software, with several partner organizations reporting bug-discovery rate gains of more than 10x after integrating AI into their testing workflows. ## The coalition… Continue reading
-
RTPurbo turns a full-attention LLM sparse in a few hundred training steps — 9.36x prefill speedup at 1M context
“Full Attention Strikes Back” introduces RTPurbo, a method that converts a standard full-attention LLM into a sparse-attention one with only a few hundred training steps — near-lossless accuracy, big efficiency gains. ## The numbers Up to 9.36x prefill speedup at 1M-token context, and about 2.01x decode speedup. The trick: keep the full KV cache only… Continue reading
-
DelTA reweights RL training so formatting tokens stop drowning out the signal that matters
DelTA is a new method for reinforcement learning from verifiable rewards (RLVR) — the training technique behind most of today’s reasoning models. The insight is sharp: the policy-gradient update in RLVR implicitly acts as a linear discriminator over token-gradient vectors, deciding which token probabilities go up or down. ## The problem it fixes That discriminator… Continue reading
-
Training Data turns microgames into a pipeline for collecting AI training data
Training Data launched on Product Hunt — an AI experience that collects training data through microgames. Players engage with small, game-like tasks, and their interactions become labeled data for training AI models. ## The data-flywheel angle The expensive input for most AI systems isn’t compute, it’s labeled data — especially human preference and interaction data.… Continue reading
-
Gated DeltaNet-2 decouples erase and write in linear attention — beats Mamba-3 and KDA at 1.3B
Gated DeltaNet-2, from the NVIDIA and MIT team behind the original, fixes a subtle flaw in how linear-attention models manage memory. Prior delta-rule models (Gated DeltaNet, KDA) used a single scalar gate to do two jobs at once — erasing old content and writing new content. v2 decouples them, and the gains show up exactly… Continue reading
-
π-Bench finds proactive assistance still stumps frontier agents — finishing a task is not the same as reducing your burden
π-Bench is a new benchmark testing something most agent evaluations skip: can an AI assistant anticipate what you need before you spell it out? It comprises 100 multi-turn tasks across 5 domain-specific user personas, and the headline finding is sobering — proactive assistance remains hard for frontier agents. ## What it tests Users rarely state… Continue reading
-
An OpenAI reasoning model disproved an 80-year-old Erdős conjecture in discrete geometry
On May 20, OpenAI announced that an internal reasoning model independently disproved a major unsolved conjecture in discrete geometry — the planar unit distance problem, first posed by Paul Erdős in 1946. ## The problem The question: how many pairs of dots can sit exactly one unit apart in the plane? For nearly 80 years,… Continue reading
Top AI Product 