gemini 3.5 flash brings agentic ai to billions
google releases gemini 3.5 flash, a fast frontier model for coding and agentic tasks, available now in search and the gemini app.
topic
ai
google launches gemini for science tools
google introduces experimental ai tools to speed up hypothesis generation, computational discovery, and literature review for researchers.
google's era ai tool writes expert scientific code
google research's empirical research assistance (era) uses gemini to automate and optimize scientific coding, now published in nature and powering the computational discovery prototype.
google io 2026 gemini agent era
google announces gemini 3.5 flash, gemini spark agent, and new ai features across products at io 2026.
daily brief: 2026-05-19
andrej karpathy joins anthropic, new rerankers and python tools arrive, and research pushes ai agents toward safety and efficiency.
python libraries for data engineering in 2026
a roundup of ten python libraries that improve data pipeline speed, reliability, and maintenance across orchestration, ingestion, quality, and storage.
andrej karpathy joins anthropic's pre-training team
ai researcher andrej karpathy moves to anthropic to work on pre-training, aiming to use claude to speed up research.
getting the most from claude cowork
claude cowork is an autonomous agent in the desktop app that works directly with your files to plan and deliver finished documents, reports, and organized folders.
light-matter particles could speed up ai chips
penn researchers created exciton-polaritons that switch signals with tiny energy, potentially enabling faster, more efficient photonic ai computing.
new ettin reranker family hits state-of-the-art at every size
six new cross-encoder rerankers from 17m to 1b parameters, built on modernbert encoders, match or beat larger models on retrieval benchmarks.
counterfactual paths cut credit assignment noise in llm reasoning
a new method uses counterfactual trajectory comparisons to turn sparse terminal rewards into step-level signals, stabilizing reinforcement learning for multi-step llm reasoning.
adversarial action removal breaks self-play rl agents
attackers who delete legal actions before an agent decides cause severe and lasting damage across multiple games and algorithms.
signmuon: 1-bit matrix-aware optimizer for distributed training
signmuon combines 1-bit sign communication with muon's matrix-aware updates to reduce distributed training bottlenecks.
skim speeds up web agents with speculative execution
skim uses site-specific templates to skip heavy ai steps for most web tasks, falling back to full agents only when needed.
agentwall adds runtime safety for local ai agents
agentwall intercepts agent actions before execution, enforcing declarative safety policies to prevent harmful operations on local machines.
stopping local ai agents early saves energy
agentstop reduces energy use in local ai agents by predicting task failure early, cutting token waste and battery drain on consumer devices.
risk-aware policy learning for offline bandits
a new method learns safe decision rules from logged data using general risk measures like cvar, with strong theoretical guarantees.
six months of llm progress in five minutes
a lightning talk summary of major llm developments from november 2025 to may 2026, including coding agent breakthroughs and open-weight model advances.