September's AI Model Wave: Anthropic, OpenAI, Meta and Google Ship Back to Back
A recap of the fastest month of frontier model releases yet, and the cyber-focused safeguards that came with them.
September 2026 brought the most compressed stretch of frontier AI releases to date. Here is how it unfolded.
The first week
In the space of a few days, all four major labs shipped updates, according to CNBC:
- Tuesday: Anthropic released Claude Fable 5.1 and Claude Mythos 5.1.
- Wednesday: Meta announced Muse Spark 1.3.
- Thursday: OpenAI launched GPT-6 Astra, emphasizing cybersecurity and computer-use skills.
CNBC noted that the Anthropic, Meta and Google releases were point updates, while GPT-6 Astra was a larger launch. The pace prompted talk of "model fatigue" among enterprise buyers trying to keep up.
Cyber safeguards move to the front
The releases came with an unusual focus on access controls. As The Hacker News reported, Anthropic's Claude Mythos 5.1 is available only through trusted-access programs, with different safeguards from the generally available Fable 5.1. OpenAI said its Astra model meets the "Critical" cybersecurity capability threshold under its Preparedness Framework, and Google is restricting its newest frontier model to vetted cyber defenders.
Later in the month
Model trackers logged further releases, including OpenAI's GPT-6 Luna and GPT-6.1 Sol and Anthropic's Claude Sonnet 5.5. October has continued the pace, with GPT-6 reaching all ChatGPT users and Anthropic's Claude Haiku 5.5 launching at a steep price cut.