OpenAI reports its researchers now consume 3.1 agent-workdays for every human workday. The company has reached its publicly stated goal of an automated research intern capable of multi-day tasks under supervision. Chief scientist Jakub Pachocki warns nobody is prepared for the coming capability jumps and calls for voluntary slowdowns and broader societal intervention.
Research acceleration: The view inside OpenAI (primary source) · An Alien Mind by Jakub Pachocki
Bryan Cantrill: readers immediately quit LLM-authored posts
Readers detect LLM prose instantly and bail. A survey of 668 developers found 78% stop reading the moment they spot it and 71% avoid the author in future. Authenticity is non-negotiable. The revolt of the reader
LiteLLM v1.100.0 adds Vertex AI Interactions and team budgets
Major release for the open-source LLM gateway brings native Vertex Interactions API support, shared team budget enforcement, numerous stability fixes, and improved CI. Release v1.100.0
Ollama 0.34 integrates local models into ChatGPT Desktop
Run open models directly inside ChatGPT Desktop from the Ollama Mac app. Also improves structured output on Apple Silicon and response compaction. Release notes
Google releases Lyria 3.5 music model in Gemini
Improved vocals and arrangements now available to all Gemini users via app and API, with new templates and longer track options. Google blog
Worth your time
Bryan Cantrill – The revolt of the reader. Essential if you publish anything online.
