<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Cognition on BriefTechNews.com</title><link>https://brief-tech-news.com/tags/cognition/</link><description>Recent content in Cognition on BriefTechNews.com</description><generator>Hugo</generator><language>en</language><copyright>Copyright &amp;copy;2021 &lt;br&gt; Designed &amp;amp; Developed by [Gethugothemes](https://gethugothemes.com/)</copyright><lastBuildDate>Mon, 05 Oct 2026 12:43:30 +0000</lastBuildDate><atom:link href="https://brief-tech-news.com/tags/cognition/index.xml" rel="self" type="application/rss+xml"/><item><title>SaaS vendors add agent tollbooths while foundation models break seat pricing</title><link>https://brief-tech-news.com/blogs/2026-10-05-founders/</link><pubDate>Mon, 05 Oct 2026 12:43:30 +0000</pubDate><guid>https://brief-tech-news.com/blogs/2026-10-05-founders/</guid><description>Salesforce, HubSpot, and Atlassian are aggressively moving away from per-seat subscriptions, hitting automated agent workflows with annual API surcharges reaching $240,000. Flat-rate pricing models are collapsing across generative AI, with legal startup Harvey recording -50% gross margins as heavy inference calls outpace fixed contracts. Meanwhile, Simon Willison makes the case for hard spend caps by default as autonomous code-generation tools create silent, five-figure cloud bills. Cognition hit $1B ARR just 24 months after launching Devin, rewriting enterprise deployment velocity.</description></item><item><title>Anthropic Splits Claude in Two, and OpenAI Admits Astra Can Hack on Its Own</title><link>https://brief-tech-news.com/blogs/2026-09-02-ai/</link><pubDate>Wed, 02 Sep 2026 13:51:41 +0000</pubDate><guid>https://brief-tech-news.com/blogs/2026-09-02-ai/</guid><description>Anthropic shipped Claude Fable 5.1 and Mythos 5.1, which are the same model with different safeguard levels: Fable is widely available at roughly 25% lower effective cost, while Mythos stays locked behind trust programs for cyber and life sciences. OpenAI says its upcoming Astra model is the first to cross its &amp;ldquo;Critical&amp;rdquo; cybersecurity threshold, meaning it finds and exploits unknown flaws without human direction, and access will be restricted to its Daybreak coalition. World Labs released Atlas, a multimodal world model that generates a minute of 1440p video from reference images using camera geometry as a native input. Cognition is raising around $1 billion at a $47 billion valuation on $900 million in annualized revenue. And an ARC-AGI-1 score of 44% now costs $0.67 and 1.5 hours of training on a single RTX 5090.</description></item></channel></rss>