<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>UNKNOWN on BriefTechNews.com</title><link>https://brief-tech-news.com/categories/unknown/</link><description>Recent content in UNKNOWN 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>Wed, 23 Sep 2026 16:07:23 +0000</lastBuildDate><atom:link href="https://brief-tech-news.com/categories/unknown/index.xml" rel="self" type="application/rss+xml"/><item><title>OpenAI’s o5 is here, and it’s 10x cheaper than o3</title><link>https://brief-tech-news.com/blogs/2026-09-23-unknown/</link><pubDate>Wed, 23 Sep 2026 16:07:23 +0000</pubDate><guid>https://brief-tech-news.com/blogs/2026-09-23-unknown/</guid><description>OpenAI dropped o5, and it’s a bigger deal than the usual model dump: the o5-mini runs at a tenth of o3’s cost, and the full o5 beats o3 on reasoning benchmarks while using less compute. Anthropic fired back with Claude 5 Opus, which edges o5 on coding and long-context. On top of that, Google’s Gemini 2.5 Pro is now free for everyone, AWS launched a $0.04-per-hour serverless GPU tier, and a critical Linux kernel privilege-escalation bug got patched in the 6.14.5 release. The model wars just got cheaper, and the infrastructure around them is following.</description></item></channel></rss>