New model launches, weights, capabilities benchmarks, model deprecations.
Anthropic's new cybersecurity-focused model, Claude Mythos Preview, may ease tensions with the Trump administration after weeks of public criticism. The release signals potential rapprochement between the AI company and government officials concerned about national security.
Alibaba open-sourced a smaller variant of Qwen3.6, signaling a strategic shift toward keeping larger, proprietary models closed to prioritize commercial revenue over community contributions.
Physical Intelligence unveiled π0.7, a robot brain model capable of performing tasks without explicit training, marking progress toward general-purpose robotic AI. The startup frames this as a meaningful advance in autonomous machine reasoning and adaptability.
OpenAI has significantly upgraded its Codex agentic coding tool with expanded capabilities, positioning it as a competitive response to Anthropic's offerings. The enhanced version grants users greater control over desktop automation and code generation tasks.
Researchers introduce MM-WebAgent, a hierarchical framework that coordinates AI-generated images and content to build visually coherent webpages while maintaining style consistency across elements. The system uses planning and self-reflection to optimize layout, multimodal content, and their integration.
SegWithU introduces a post-hoc uncertainty quantification framework for medical image segmentation that operates in a single forward pass by modeling uncertainty as perturbation energy in a compact probe space, enabling both calibration and error detection without repeated inference.
Meituan researchers propose Policy-Guided Hybrid Simulation (PGHS), a dual-process framework combining LLM reasoning with learned behavioral policies to simulate merchant-level user behavior for counterfactual strategy evaluation without costly online experiments.
Researchers propose MambaSL, a single-layer Mamba variant optimized for time series classification, achieving state-of-the-art results across 30 UEA datasets. The work also re-evaluates 20 baseline models under unified benchmarking protocols to address reproducibility gaps in the field.
Researchers introduced QuantCode-Bench, a 400-task benchmark for evaluating LLMs on generating executable algorithmic trading strategies for the Backtrader framework. The benchmark tests whether models can combine financial domain knowledge, API mastery, and correct syntax to produce strategies that execute on historical data.
Researchers simulated cortical development from a single stem cell using gene regulatory rules, generating 85 mature neurons that spontaneously self-organized into a 200k-synapse circuit. The minimal network jumped from chance-level MNIST performance to 89–94% accuracy after one training epoch, demonstrating how developmental constraints can yield efficient learning architectures.
Researchers introduce LAMAE, a masked autoencoder foundation model designed for multi-view echocardiography that uses latent attention to share information across cardiac imaging frames and views. The approach addresses limitations of frame-independent processing by enabling coherent reconstruction of heterogeneous spatiotemporal cardiac data.
Researchers introduced ProVoice-Bench, a new evaluation framework for proactive voice agents with 1,182 test samples across four novel tasks. Testing state-of-the-art multimodal LLMs revealed significant performance gaps, particularly in over-triggering and reasoning, exposing limitations in current models' ability to anticipate and intervene proactively.
Researchers benchmarked six multilingual embedding models (Potion, Gemma, BGE, Snow, Jina, E5) for hate speech detection across Lithuanian, Russian, and English using a new Lithuanian corpus (LtHate) and existing datasets, comparing anomaly detection and classification approaches.
Researchers introduce DynAfford, a benchmark for testing embodied AI agents in environments where object affordances shift dynamically and aren't explicitly stated. The accompanying ADAPT module helps planners infer hidden preconditions and adjust actions in real-world scenarios where naive instruction-following fails.
OpenAI unveiled GPT-Rosalind, a specialized reasoning model designed to enhance drug discovery, genomics analysis, and protein research workflows in life sciences. The model targets scientific researchers seeking to accelerate computational biology and pharmaceutical development pipelines.
Google DeepMind released Gemini 3.1 Flash TTS, an audio model featuring granular audio tags that enable fine-grained control over expressive speech synthesis. The capability allows developers to direct AI-generated audio with unprecedented precision for creative and commercial applications.
OpenAI expanded its Trusted Access for Cyber program by releasing GPT-5.4-Cyber, a specialized model for vetted cybersecurity professionals, while implementing enhanced safeguards to manage risks from advanced AI security capabilities.
Google DeepMind released Gemini Robotics-ER 1.6, an embodied reasoning system designed to improve spatial understanding and multi-view perception for autonomous robotic systems. The update enhances the model's ability to reason about physical environments and execute real-world manipulation tasks.