Latest stories
EmbodiedMemory-Bench Tests Whether Embodied Agents Can Actually Remember
A new benchmark of 2,554 interactive episodes probes four memory weaknesses in long-horizon embodied agents, alongside a structure...
Adaptive Looped Transformers Boost Test-Time Scaling by 53% on AIME
A new paper introduces TaH2, a post-training method that directs extra loop iterations only to tokens that benefit from them, impr...
Hugging Face Paper Proposes 'Physical Coding' to Make Robots Learn Like Software Agents
A new paper argues that representing robot task state and execution as code can overcome the brittleness of vision-language-action...
TraceDance Turns Real Agent Deployment Traces Into Targeted Behavior Benchmarks
Hugging Face paper TraceDance automatically builds benchmarks from real agent deployment traces to test specific undesirable behav...
YuE2 Unifies Symbolic and Audio Music Generation, Rivaling Suno
Hugging Face researchers introduce YuE2, a single Mixture-of-Transformers model that plans music as a readable score before render...
Anthropic Unveils Claude Sonnet 5.5: Faster, Cheaper, and Built for Everyday Work
Anthropic's new Claude Sonnet 5.5 runs over 30% faster than Sonnet 5 and costs up to 30% less per task, while posting dramatic gai...
Manus 2.0 Arrives: New Agent Architecture, Studio Suite, and a Standalone App Called Cue
Manus has launched Manus 2.0, describing it as a new architecture rather than a version update, spanning a leaner agent harness ca...
OpenAI Expands Lenfest AI Collaborative with $5M Funding and $5M in Credits
OpenAI is expanding the Lenfest AI Collaborative and Fellowship Program with $5 million in funding plus up to $5 million in softwa...
Manus Resumes Independent Operations as a Standalone Agent Lab
Manus says it is resuming independent operations, positioning itself as a standalone agent lab focused on pushing the boundaries o...
SLCA-GRPO Fixes Credit Misattribution in Tool-Calling RL
A new Hugging Face paper introduces SLCA-GRPO, a framework that locks credit assignment to structural segments so summary rewards...
TrackEverything Tracks Every Point in 3D Across 1000+ Frame Videos
A new 3D point tracker from Hugging Face researchers represents videos as persistent 3D scene tracks in world coordinates, breakin...
SAGE Framework Targets Long-Horizon Reasoning Biases in LLMs
A new paper introduces Symbolic Closure Analysis and the SAGE framework, which uses algebraic sparsification and hyperbolic struct...