AI roundup
AI Weekly Roundup — Sep 26, 2026
Sat 26 September 2026
Model releases
Meta Steals Spotlight From OpenAI and Anthropic During Model Drop Week
Anthropic rolled out Opus 5.5, only to be followed by OpenAI's GPT-6 model updates roughly 90 minutes later in a compressed model-drop week. Meta nonetheless stole the spotlight from both frontier labs.
Source: TechCrunch AI
Anthropic Releases Opus 5.5
Anthropic released Opus 5.5 on Tuesday. The excerpt provided no additional technical or pricing details.
Source: Zvi - Don't Worry About the Vase
Anthropic Says Claude Opus 5.5 Cuts Token Costs by ~40%
Anthropic estimates that Claude Opus 5.5 costs about 40% less to run than Opus 5 for typical workloads billed by token, with the greatest savings on longer-running, higher-context workloads. The pitch is aimed at developers watching inference spend.
Source: Anthropic (community mirror)
Anthropic, xAI, and Xiaomi Drop Models Within 24 Hours
Anthropic released Claude Opus 5.5 today, following yesterday's rollout of xAI's Grok 4.7 and Xiaomi's MiMo v2.6 Flash/Pro. The back-to-back releases made for a crowded two-day window in the model release calendar.
Source: Simon Willison
Industry & funding
Anthropic Commits $11.6B to Akamai, With Equity Kicker
Anthropic will pay Akamai $11.6 billion over seven years for cloud infrastructure, a figure that could rise to roughly $20 billion. As part of an unusual arrangement, Akamai will grant Anthropic an equity stake of up to 5% that scales with its spending.
Source: TechCrunch AI
Nscale Lands Funding From Third Point and Nvidia for Data Center Buildout
Nscale has pulled in funding from Third Point, Nvidia, and others to expand its AI data center footprint. The excerpt did not specify the amount or terms of the financing.
Source: TechCrunch AI
Anthropic Founders Seek 50.1% Voting Control
Anthropic is asking shareholders to approve a structure that would grant its seven co-founders a combined 50.1% voting stake on most corporate matters. The arrangement would tightly concentrate governance power in the founding team.
Source: TechCrunch AI
Australia Investigates Whether OpenAI Broke the Law Over Hack
Australia's prime minister expressed disappointment at being informed of a hack solely via email, and the country is now investigating whether OpenAI violated the law. The excerpt did not detail the target or the method of the breach.
Source: Wired AI
Tools
Meta Pushes Muse AI Agent as It Tops App Store Charts
Muse is topping app store charts and adding users quickly as Meta ramps up promotion of the personal AI agent across its own apps and beyond. The traction gives Meta a rare consumer-AI win driven by its existing distribution network.
Source: TechCrunch AI