Yu Fu

It is suggested that safety does not automatically scale with reasoning capability, especially under long-context inference, as context length increases and safety alignment consistently degrades.

StepAudio 3 Realtime, an audio-language foundation model organized around a continuous listen-converse-think-act loop, achieves dialogue and reasoning performance comparable to dedicated reasoning models while speaking in real time, and resolves the tension between deep deliberation and latency via Think-While-Speaking.

This study introduces StepAudio 3 Gen, a general-purpose audio generation model that supports zero-shot text-to-speech (TTS), voice design, vocal generation, sound effects, music, vibe speech, and mixtures of multiple audio types within a unified framework.

Frontis-MA1 achieves a remarkable 71.21% Medal Average on MLE-Bench Lite, outperforming leading models and showcasing the potential of AI systems to recursively improve their own engineering processes.

PhaseAware reduces rehabilitation scoring error by nearly 89% while providing interpretable cues for clinician review, bridging automation and human oversight.

Data narratives often fail not just due to errors, but through a complex interplay of meaning-making processes that can mislead audiences.

A 3B model, guided by a novel RL framework, can outperform a 20B model in capturing diverse human perspectives, challenging the assumption that larger models inherently possess better alignment.