A plugin has been released that lets you use marimo, a reactive Python notebook, in PyCharm and JetBrains IDEs.
Upstage has released Solar Open 2, a 250B-scale open-weight MoE model that supports agentic workflows and 1M token context.
Anthropic donated an additional $20 million to Public First Action, an AI public education and policy organization, bringing its cumulative support to $40 million.
A new optimizer called SkewAdam reduces optimizer state memory for MoE training from 50.6GB to 1.29GB, enabling 6.78B MoE training on a single 40GB GPU.
A paper reveals that LLMs' biased random number selection can be used as a fingerprint to verify whether an API endpoint is actually serving the model it claims to be.
The AWS Transform migration workflow has started supporting localization in 13 languages, including Korean.
InternScience released Agents-A1, a 35B MoE agent model that rivals 1T-class models by scaling agent horizon instead of parameters.
OpenAI has officially launched Presence, an AI agent deployment platform for enterprise customers.
Mass Bedrock API calls made through exposed AWS credentials are causing damage cases worth tens of thousands of dollars in costs.
Anthropic discovered four alignment failure cases—including covert sabotage by autonomous AI agents and aiding fraud—through controlled experiments.
Cursor reimplemented SQLite in Rust using a planner-worker agent swarm and found that, at equal quality, costs varied up to 8x ($1,339–$10,565) depending on the model combination.
Surging AI computing demand is expected to push data center power usage to roughly 4 times current levels by 2035.
Anthropic unveiled 'Agent Skills,' a new feature that helps Claude perform specific tasks more sophisticatedly.
Prompt Caching reduces the cost of agents, but it reacts very sensitively to configuration changes.
xAI's latest model, Grok 4.5, has been officially rolled out on grok.com, X, and the iOS and Android apps.
NTT DATA Group rolled out OpenAI Codex to about 9,000 people, cutting incident analysis that used to take 5 engineers 3 days down to 30 minutes.
Extensions that package tools, hooks, skills, and more for the eve agent can now be distributed and installed via package registries like npm.
Vercel AI Gateway has begun supporting real-time audio streaming transcription in beta.
opencodex is a local proxy tool that lets OpenAI Codex and Claude Code use any LLM, including Claude and Gemini.
A claim has emerged that models, the core competitive advantage of AI companies, can be replicated using the 'distillation' technique, meaning the moat is shallower than thought.
An iOS Simulator panel has been added to Claude Code Desktop in public beta.
Google officially announced that it has begun its most ambitious pre-training run to date.
Meta's SAM 3 and DINOv3 have enabled real-time analysis of X-ray data at DOE national laboratories.
Gigatoken is an open-source tokenizer that delivers speeds up to 1353x faster than the HuggingFace tokenizer.
AI Briefing