An international research team released the largest video reasoning dataset to date—roughly 1,000 times larger than previous alternatives. Testing reveals that state-of-the-art models including Sora 2 and Veo 3.1 substantially underperform humans on reasoning tasks, suggesting the limitation isn't data scarcity but architectural constraints.
Researchers introduce FlyThinker, a framework that runs reasoning and generation concurrently rather than sequentially, addressing limitations of existing "think-then-generate" approaches in long-form personalized text generation. The method uses a separate reasoning model that generates token-level guidance in parallel with the main generation model, enabling more adaptive reasoning without sacrificing computational efficiency.
Researchers have introduced OmniVideoBench, a large-scale evaluation framework comprising 1,000 manually verified question-answer pairs derived from 628 videos (ranging from seconds to 30 minutes) designed to measure synergistic audio-visual reasoning in multimodal large language models. Testing reveals a significant performance gap between open-source and closed-source MLLMs on genuine cross-modal reasoning tasks.
Researchers propose TTSR, a test-time training framework where a single LLM alternates between Student and Teacher roles to improve its own reasoning. The method generates targeted variant questions based on analyzed failure patterns, showing consistent improvements across mathematical reasoning benchmarks without relying on unreliable pseudo-labels.
Researchers introduced T2S-Bench, a new benchmark with 1,800 samples across 6 scientific domains and 32 structural types, evaluating text-to-structure reasoning in 45 mainstream models. The benchmark reveals substantial capability gaps: average accuracy on multi-hop reasoning tasks is only 52.1%, while Structure-of-Thought (SoT) prompting alone yields +5.7% improvement on average across eight text-processing tasks.
A new geometric framework models how large language models reason through embedding trajectories that evolve like physical flows. Researchers tested whether LLMs internalize logic beyond surface form by using identical logical propositions with varied semantic content, finding evidence that next-token prediction training leads models to encode logical invariants as higher-order geometry.
A new training approach using knowledge graphs as implicit reward models enables a 14-billion-parameter model to outperform much larger systems like GPT-5.2 and Gemini 3 Pro on complex multi-hop reasoning tasks. Researchers combined supervised fine-tuning and reinforcement learning with knowledge graph path signals to ground models in verifiable domain facts.
A new geometric framework demonstrates that LLMs internalize logical reasoning as smooth flows—embedding trajectories—in their representation space, rather than merely pattern-matching. The research, which tests logic across different semantic contexts, suggests next-token prediction training alone can produce higher-order geometric structures that encode logical invariants.
Researchers introduced CareMedEval, a 534-question benchmark derived from French medical student exams, to evaluate LLMs on biomedical critical appraisal and reasoning tasks. Testing state-of-the-art models reveals none exceed 50% exact match accuracy, with particular weakness in evaluating study limitations and statistical analysis.
Researchers present REFLEX, a framework that equips LLM-powered robotic agents with metacognitive capabilities—skill decomposition, failure reflection, and solution synthesis—to perform complex tasks in zero-shot and few-shot settings. The system significantly outperforms existing baselines and demonstrates that LLMs can generate creative solutions that diverge from ground truth while still completing tasks successfully.
Researchers propose LaDiR, a framework that replaces traditional autoregressive decoding with latent diffusion models to improve LLM reasoning. The approach encodes reasoning steps into compressed latent representations and uses bidirectional attention to refine solutions iteratively, enabling parallel exploration of diverse reasoning paths.
Researchers introduced CFE-Bench (Classroom Final Exam), a multimodal benchmark using authentic university homework and exam problems across 20+ STEM domains to evaluate LLM reasoning capabilities. Gemini 3.1 Pro Preview achieved the highest score at 59.69% accuracy, while analysis revealed frontier models frequently fail to maintain correct intermediate states in multi-step solutions.
Researchers have released HSSBench, a new benchmark designed to evaluate multimodal large language models on humanities and social sciences tasks—areas where current benchmarks are sparse. The benchmark contains over 13,000 samples across six key categories in multiple languages, and testing shows even state-of-the-art models struggle significantly with cross-disciplinary reasoning required for HSS domains.
Researchers propose Perception-R1, a method that adds visual perception reward signals to reinforcement learning training for multimodal AI models. The approach achieves state-of-the-art results on multiple reasoning benchmarks using just 1,442 training examples by explicitly teaching models to accurately perceive visual content before reasoning about it.
Researchers propose LaDiR (Latent Diffusion Reasoner), a framework that combines variational autoencoders and latent diffusion models to improve LLM reasoning. The approach encodes reasoning steps into continuous latent representations, enabling iterative refinement and parallel generation of diverse solutions beyond traditional autoregressive decoding.
DeepSeek has released DeepSeek-R1, a text generation model featuring reasoning capabilities through chain-of-thought processing. The model was published January 20, 2025 and has accumulated over 830,000 downloads on Hugging Face.
Inception has released Mercury 2, positioning it as the first diffusion-based language reasoning model. Rather than generating text sequentially word-by-word like standard language models, Mercury 2 refines entire passages in parallel, according to the company.
Google announced Gemini 3.1 Pro, positioning the model for complex problem-solving tasks requiring deeper reasoning than previous versions. The release follows Gemini 3 Pro (November 2025) and Gemini 3 Flash (December 2025).