Reasoning Models Ace the CFA Exams

Google has introduced a new benchmark called 'FACTS' aimed at measuring the factual accuracy of generative AI models, addressing a critical gap in existing benchmarks that focus primarily on task completion rather than the truthfulness of the information generated. This initiative is particularly significant for industries where accuracy is essential, such as legal, finance, and medical sectors.

A new dataset for Automatic Essay Scoring (AES) and feedback generation in Basque has been introduced, consisting of 3,200 essays annotated by experts. This dataset targets the CEFR C1 proficiency level and includes detailed feedback on various scoring criteria. The study demonstrates that fine-tuning open-source models like Latxa can outperform established systems such as GPT-5 in scoring consistency and feedback quality.

A new framework named ReasonBreak has been introduced to address privacy concerns associated with multi-modal large reasoning models (MLRMs), which can infer precise geographic locations from personal images using hierarchical reasoning. This framework employs concept-aware perturbations to disrupt the reasoning processes of MLRMs, aiming to enhance geographic privacy protection.

A recent study has demonstrated the potential of large language model (LLM) agents to automate high energy physics data analysis, specifically using the Higgs boson diphoton cross-section measurement as a case study. This hybrid system integrates an LLM-based supervisor-coder agent with the Snakemake workflow manager, allowing for autonomous code generation and execution while ensuring reproducibility and determinism.

OpenAI has launched upgrades to its GPT-5 model, introducing GPT-5.1 Instant for improved instruction following, GPT-5.1 Thinking for faster reasoning, and GPT-5.1-Codex-Max for enhanced coding capabilities. These updates aim to enhance user interaction and response quality in AI applications.

The introduction of PersonaMem-v2 marks a significant advancement in AI personalization, featuring a dataset that simulates 1,000 user-chatbot interactions across diverse scenarios, revealing user preferences implicitly. This dataset aims to enhance long-context reasoning capabilities in AI models through reinforcement fine-tuning, addressing the challenges faced by current large language models (LLMs) in achieving effective personalization.