Microsoft Research Finds Clearing Chat History Helps AI Agents Improve
Microsoft researchers found AI agents can achieve higher scores by clearing chat history, with Fork-and-Flush boosting performance by 66% on the same compute budget.
Archive
Browse all published articles.
ARCHIVE
-
-
Meta Releases Muse Wonder Full, Enters AI Top Tier
Meta's Muse Wonder Full scores 56 on the Artificial Analysis Intelligence Index, surpassing mid-tier status and entering the AI top tier.
-
Amazon Expands Robotic Workforce in Warehouses
Amazon operates over 1 million robots in warehouses, with plans to significantly increase the fleet as it builds new factories.
-
Ukraine strikes Yandex data centers in Russia's digital war
Ukraine’s drone strikes have disabled three of Russia’s largest data centers, including those housing Yandex’s supercomputers used for AI training.
-
Blueprints AI Unveils AI-Driven Blueprint Generator for Construction
Blueprints AI, a startup founded by Al Dram, announced it can generate permit-ready blueprints in minutes using AI, after six months of development following ChatGPT’s release in 2022.
-
Microsoft President Calls for Urgent AI Regulation
Microsoft president Brad Smith urged global AI safety rules, citing the U.K.'s leading role in testing major models like those from OpenAI and Anthropic.
-
WindBorne Secures $50M Air Force Contract for Balloon Data
The U.S. Air Force has awarded WindBorne Systems a contract worth up to $49 million for data collected by its global network of weather balloons, marking the company's largest deal to date.
-
IBM Unveils Distributed Inference Platform for AI Workloads
IBM announced a new distributed inference platform in September 2026, designed to handle complex AI tasks with improved efficiency.
-
Amazon Bedrock Adds Global Cross-Region Inference for OpenAI GPT-6 Models
Amazon Bedrock now allows global cross-region inference for OpenAI GPT-6 models in Thailand, Malaysia, Singapore, Indonesia, and Taiwan, enabling seamless access to high-capacity model deployments.
-
Hugging Face Introduces LeRobot Dataset Quality Metrics
Hugging Face released new LeRobot dataset quality metrics to help users evaluate robot training data, with 6 key questions to identify issues before training.
-
HuggingFace Releases ColGrep-Agent-2B Code-Search Model
HuggingFace announced ColGrep-Agent-2B, a 2-billion-parameter code-search model that reduces API tokens by 34% in testing.
-
Noitom Robotics Introduces Adaptive Motion Planning for Humanoid Robots
Noitom Robotics outlines an adaptive motion planning framework for humanoid robots, validated on real-world platforms with no motion-capture systems.