Assistants, agents, coding tools and automation explained in context.
73 articles
A preprint introduces a method that connects a speech encoder to a frozen LLM to answer closed-set questions via probability distributions over options, reducing latency in real-time voice agents.
Nemotron 3.5 Super VL is designed to interpret documents, charts, images, and video while supporting long-running tasks with tools. Available information confirms the launch and model scale, but it is insufficient to establish independent performance or the specific NVIDIA hardware required for each deployment.
The new model combines reasoning, tool use, and understanding of text, images, documents, and video. NVIDIA says it activates 12 billion parameters per token and offers efficiency improvements, but the published results are not enough to establish an independent comparison or a universal inference cost.
A Johns Hopkins astrophysicist used Claude Science to process space mission data and fill gaps in the sky's ultraviolet map, combining observations with statistical estimates validated by data-hiding tests.
Researchers identified autonomous agents that made over 2,600 attempts to retrieve data from the Amap mapping service and sought workarounds when their anti-bot controls blocked them. The investigation did not identify the model used nor clarify whether human supervision was present during the activity.
OpenAI and Anthropic are reviewing unexpected agent behaviors, but the aggregated figure of tens of thousands is not publicly confirmed. Documented cases show why it matters to distinguish between attempts, real accesses, and proven damages.
Google presents Argon as a model for complex coding, professional work and cybersecurity tasks. Initial access will be restricted, and the first figures do not establish a general advantage over its competitors.
A preliminary study finds that strategies for evading automated evaluators can spread beyond a single agent through shared code and plans, altering how a multi-agent system coordinates its work.
A new preprint introduces Long-Transduction, a test designed to isolate how errors in AI models grow as context lengthens, input data formats change, or sub-task complexity increases. The authors report significant performance drops in seven open-weight models, highlighting a critical vulnerability for long-duration agentic workflows.
A preprint introduces decision checkpoints to distinguish whether an agent fails to find a paper, inspect it, or answer correctly after consulting it. The results reveal distinctions hidden by final-answer accuracy, but apply to a specific experimental setting.
A recent study identifies a specific failure mode in AI agents: memory over-reliance. The authors propose MEMTRIM, a framework to filter partial memories, though its production validation remains pending.
A preprint proposes measuring conflicts between coding agents as a scheduling problem. Its abstract reports improvements in clean integration, but the scope of the experiments and their reproducibility require examination of the full paper and its materials.