Together AI presented eight papers at ICML 2026 spanning the full AI stack, from frontier agents to GPU kernels. Their research includes DSGym for standardized data-science agent evaluation, ThunderAg…
Google researchers introduced the Paper Assistant Tool (PAT), an AI framework that automates scientific review by evaluating manuscripts for errors and suggesting improvements. PAT achieved a 34% impr…
A developer built paperlist.ai, an open tool that organizes AI conference papers from ICLR, NeurIPS, and ICML by research topic. The tool uses automated topic assignment and manual curation to help re…
Researchers from Peking University developed ECG-R1, a multimodal vision-language model trained with reinforcement learning to differentiate ischemic from nonischemic T-wave inversion on ECGs. The mod…
Researchers from KAIST introduced Video-based Optimal Transport Preference (VOTP), a method that uses optimal transport over Video Foundation Model embeddings to generate high-fidelity pseudo-labels f…
A Peking University team presented Meta Context Engineering (MCE) at ICML 2026, achieving an 89.1% average improvement over a base agent across five domains, outperforming the previous best method (AC…
Swiss AI researchers released the Apertus Mini collection, 16 small language models distilled from the Apertus v1 8B model, available in 0.5B, 1.5B, and 4B parameter sizes with multiple quantization l…
On June 11, 2026, OpenAI committed to a 10-gigawatt data center, marking a shift toward national-scale AI infrastructure, while a separate study showed that finetuning a model on a book's plot alone r…
A small study testing whether wrapping untrusted prompt inputs in mock tool calls could improve language model robustness found the technique did not broadly help across three LLM-as-a-Judge tasks and…
Researchers from NVIDIA have developed a new sparse data format and custom GPU kernels, called TwELL, that reshape unstructured sparsity in transformer language models to align with GPU architecture, …