#multi-step reasoning

'Tutorial shows how to assemble a real-time voice AI agent that transcribes, reasons, plans and speaks using Whisper and SpeechT5.'

Fractional Reasoning introduces a model-agnostic method to adaptively control reasoning depth in LLMs, enhancing performance and efficiency on complex reasoning tasks.

The Deep Research Bench report by FutureSearch evaluates AI agents on complex research tasks, revealing strengths and key limitations of leading models like OpenAI's o3 and Google Gemini.

Apple and Duke researchers introduce Interleaved Reasoning, a reinforcement learning method that allows LLMs to produce intermediate answers, significantly boosting response speed and accuracy in complex tasks.

Dream 7B introduces a diffusion-based reasoning approach that enhances AI's ability to reason, plan, and generate coherent text, outperforming traditional autoregressive models.

Microsoft launched the Phi-4-Reasoning family, a set of 14B parameter open-weight models optimized for complex reasoning tasks. These models demonstrate competitive performance on math, planning, and coding challenges with transparent training and open access.