Anna Goldie: the brain is still more efficient than AI
Ricursive CEO Anna Goldie told The Information the brain, at about 20 watts, is still more efficient than AI. Her thesis is chip design, not a brain chip.
Models
Model cards, releases, evaluations, and the limits a lab states in public.
Dense Qwen3.8-27B now on Nebius Token Factory
On 30 Sep 2026 Qwen said Qwen3.8-27B is on Nebius Token Factory. Weights shipped 14 Aug. Price and speed come from the card and third parties.
Perplexity ships pplx-embed-v2-context-9b-preview
Perplexity released pplx-embed-v2-context-9b-preview on 30 Sep 2026. Weights are on Hugging Face; SOTA scores are the company's own.
Microsoft joins Apache Ossie, an open metrics standard
At FabCon Europe, Microsoft confirmed Apache Ossie and published a Power BI converter. The standard records metrics; complex DAX still becomes BLANK().
DeepMind open-sources SynthID Bio protein watermark
Google DeepMind published SynthID Bio in Nature on 30 Sep 2026: open-source tools to watermark AI proteins. Lab tests are the company's own.
Google pays some publishers for AI answers
Google confirmed a closed Search Console pilot that pays some sites for AI Overviews, AI Mode and Gemini. List, formula and sums stay unpublished.
Google launches Gemini 4 Argon for defenders first
Google DeepMind announced Gemini 4 Argon on 30 September 2026. Fairwind defenders get it first, without cyber guardrails. No public API date yet.
Dazzle AI uses camera roll as assistant context
Marissa Mayer's Dazzle AI launched an assistant that builds context from photos and screenshots, not email. Invite-only iOS, SMS, and web.
AI data centers need power faster than the grid
IEA and Rystad say AI data centers need power faster than the grid. On-site gas and fuel cells fill the wait. The Information financing angle is a RUMOR.
Cursor visits Innate, a 7-person robot team
Cursor posted a visit video to Innate, a YC Fall 2024 personal robot startup. The clip is marketing, not independent reporting.