A groundbreaking library emerges, enabling scalable large language model (LLM) inference through flexible tree search, featuring the AB-MCTS algorithm, customizable node generation and scoring, checkpointing, search resumption, and multi-LLM support, revolutionizing efficient LLM deployment.
In a clash of AI strategies at Meta, Zuckerberg launches Meta Superintelligence Labs to aggressively pursue superintelligence development, contrasting with Chief AI Scientist LeCun's cautious, open-source approach focused on 'world models' and animal-level intelligence.
Sakana AI unveils 'AI Dream Team' technique enabling multiple large language models to cooperate, combine strengths, and solve complex problems through trial-and-error, releasing open-source TreeQuest framework for robust enterprise AI systems.
Google unveils Gemini CLI, a new coding agent that trails Claude Code from Anthropic in speed, cost, token efficiency, and user experience, despite delivering decent code quality.
Meta CEO Mark Zuckerberg denies reports he offered top AI talent from OpenAI massive compensation packages worth up to $300 million, with up to $100 million vesting in the first year, as Meta continues to aggressively hire AI researchers from OpenAI.
In a bold move, Meta CEO Mark Zuckerberg unveils Meta Superintelligence Labs, poaching top AI talent from rivals like OpenAI and Google to spearhead development of cutting-edge foundation models and AI products under the new division.
Researchers unveil Hunyuan-A13B, a groundbreaking 80B-parameter open-source AI model that employs a fine-grained Mixture-of-Experts architecture, delivering competitive performance while optimizing efficiency and scalability across various benchmarks.
Warp achieves a remarkable 52% success rate on Terminal-Bench, outperforming rivals by 20% and securing the #1 spot through an optimal model fallback chain, agent control over long commands, and enforced todo list maintenance, utilizing Claude Sonnet 4 as primary and Claude Opus 4 as planning model.
Large Language Models are accelerating exponentially, with capabilities doubling every 7 months, reshaping software development by 2030 as advanced LLMs could complete month-long tasks with 50% reliability, often much faster, raising significant potential benefits and risks.
Baidu unveils a groundbreaking multi-agent AI search system that utilizes a Directed Acyclic Graph and specialized agents to intelligently break down complex queries, execute subtasks, and synthesize enhanced information retrieval responses.