Google says its September 2026 AI updates include a new frontier model, Gemini 4 Argon, alongside Gemini 3.8 models for software engineering, cybersecurity and voice interactions. In its October 2 roundup, the company describes restricted access for its cybersecurity offerings and a broader rollout for audio tools.
Gemini 4 Argon starts with cybersecurity defenders
Google describes Gemini 4 Argon as its newest frontier model, with a 1-million-token context window. The company says it is designed for coding, cybersecurity defense, financial research, legal drafting and autonomous cybersecurity patching.
Access remains limited in Google’s account: Argon is rolling out to trusted cyber defenders through the Fairwind Program. Google says it is gathering feedback from early testers to refine guardrails before expanding availability to developers, enterprises and consumers. That makes the announcement a restricted rollout with planned expansion, rather than evidence of general developer access.
Flash updates target software engineering and patching
Google says Gemini 3.8 Flash improves long-horizon software engineering, multi-step problem solving and agentic workflows. It also claims improved reasoning and coding performance while maintaining the speed and low cost of Gemini 3.7 Flash. Those performance comparisons are company claims; the supplied evidence does not include independent benchmark results or numerical pricing details.
For security teams, Google describes Gemini 3.8 Flash Cyber as offering autonomous vulnerability discovery and automated code patching. The company says this model is available to trusted organizations through the Fairwind Program. Both cybersecurity announcements therefore tie access to that program, even as Google outlines broader ambitions for Argon.
Voice models and Connected Apps
Google says it launched Gemini 3.8 Live for everyday conversation and rolled out both Live and Live Extended Thinking to developers through Google AI Studio and the Gemini API. The company also lists Gemini 3.8 Live availability across Search Live, Gemini Live and Google Workspace, with voice features in Gmail, Docs and Keep.
The roundup introduces Gemini 3.8 Flash-Lite TTS and Gemini 3.8 Flash TTS as text-to-speech models. According to Google, they support generating custom voices and directing scene dialogue from text prompts. Google says Flash TTS is rolling out across products including Gemini Notebook and Google Vids.
Google also lists new Connected Apps in Gemini: Airtable, Linear, monday.com and PandaDoc for productivity; Adobe, Picsart, Squarespace and Webflow for creative work; and Peloton, SeatGeek and Experian among lifestyle services. The supplied evidence names these integrations without detailing their individual permissions or supported actions.
The capability and availability descriptions here come from Google’s announcement. They have not been independently corroborated in the supplied research, and the planned expansion of Argon should be read separately from the access Google says is already rolling out.
