In partnership with
Welcome back to The AI Field. Two weeks after a cyber-first preview, Google has opened Gemini 4 Argon to paying users and developers, at a per-token price well below its main rivals. It arrives one day after Meta's Muse Wonder, so the top of the model race just got crowded again.
Inside: who can use Argon today and what it costs, how OpenAI says its own models helped catch an engineer accused of bugging an executive meeting room, and Apple's HomeView, the first device built around Siri AI. Plus six tools to try and a ten-minute test to see if Argon beats your current model on real work.
In today's AI Field:
Gemini 4 Argon opens to paying users and developers
An OpenAI engineer is charged over a hidden recorder
Apple's HomeView brings Siri AI into the home
Plus: Shopify lets any agent shop its catalog, Amex adds a purchase check for agents, and Deezer lets you filter out AI music.
Read time: 5 minutes
1,000+ Proven ChatGPT Prompts That Help You Work 10X Faster
ChatGPT is insanely powerful.
But most people waste 90% of its potential by using it like Google.
These 1,000+ proven ChatGPT prompts fix that and help you work 10X faster.
Sign up for Superhuman AI and get:
1,000+ ready-to-use prompts to solve problems in minutes instead of hours—tested & used by 1M+ professionals
Superhuman AI newsletter (3 min daily) so you keep learning new AI tools & tutorials to stay ahead in your career—the prompts are just the beginning
LATEST DEVELOPMENTS
GOOGLE · MODELS
The AI Field: Google has made Gemini 4 Argon, its new frontier model, available worldwide through the Gemini API and to Google AI Pro and Ultra subscribers in the Gemini app and AI Mode in Search. Until now, only vetted cyber defenders in Google's Fairwind Program could use it.
The details:
The API launches at an introductory $2 per million input tokens and $10 per million output tokens, with cached input 95% off. After November 30 the price doubles to $4 and $20, which is still below OpenAI's and Anthropic's top models per token.
Google says Argon ranks #1 on the Vals Index of finance, legal, coding and tax work, and cut prompt-injection attack success to 0.7% on the Gray Swan benchmark. Inside Google, teams using it in Antigravity shipped over 35% more agent-written code with nearly a third fewer rollbacks.
The Gemini app update adds a chat-to-tasks toggle, links to more than 40 apps like Instacart, OpenTable, Adobe and Shopify, and puts Antigravity's coding interface inside the Mac app. Free and Plus users don't get Argon yet, and there is no date for when they will.
Why it matters: Google spent months trailing OpenAI and Anthropic, and Argon is its first real frontier answer since Gemini 3. It is priced to win switching tests. The benchmarks are Google's own picks, so the useful question is narrower: does it handle your hardest recurring task better or cheaper than what you use now? If you are on the API, the intro price until November 30 is a cheap window to find out.
OPENAI · SECURITY
The AI Field: Federal prosecutors have charged an OpenAI engineer with planting a covert audio recorder in an executive conference room at the company's San Francisco offices. OpenAI says it used its own models to analyze the device and trace it to the employee, then referred the case to the FBI.
The details:
According to the complaint, the encrypted device was hidden in toilet paper and plugged into a USB port behind a TV around September 28, then found two days later. A second, similar device turned up in a bag he left unattended on October 5.
Abdulrahman Alfozan, 35, a Saudi national who joined OpenAI in July 2025, was arrested on October 9 and appeared in federal court on Tuesday. He faces up to five years in prison if convicted.
Prosecutors say a chatbot account linked to his spouse asked how to build, encrypt and remotely access a continuous audio recorder in the days before. OpenAI says it has no sign he accessed or took model weights, source code or customer data.
Why it matters: AI labs now guard their meeting rooms the way defense contractors do, and the US government is treating AI secrets as national security. For everyone else the lesson is more ordinary. Insider risk is physical as well as digital, and prompt history can end up as evidence. Check what your own AI tools log, how long they keep it, and who on your team can see it.
Leave Granola and get up to 12 months free of Wispr Flow Notetaker + Dictation
If you have paid time left on an individual Granola plan, we'll match it with a Wispr Flow subscription that includes Notetaker and dictation, and add bonus time, up to 12 months total. Sign in or create a Wispr account and submit proof of your plan to check eligibility.
APPLE · DEVICES
Image: Apple
The AI Field: Apple has unveiled HomeView, a $299 smart display with a 7-inch screen built around Siri AI, its rebuilt conversational assistant. Pre-orders are open in the US, UK, Canada, Australia, India and New Zealand, with deliveries from November 13.
The details:
HomeView recognizes up to six people by face and shows each person their own messages, calendar and reminders. With an Apple Intelligence iPhone or iPad nearby, Siri AI can take actions across apps using personal context.
Requests run on the device or on Apple's Private Cloud Compute. Siri AI is labeled beta on HomeView, and some server-side features have daily limits, with expanded access coming "for a fee" later.
It won't be sold in the EU because Siri AI isn't available in iOS there yet. Apple also announced a new HomePod mini and Apple TV 4K, plus cameras, thermostats and a lock built with LG and Schneider Electric.
Why it matters: This is the first Apple device sold mainly as a showcase for the new Siri, and it goes straight at Amazon's and Google's home screens. Read the fine print: heavier AI use will cost extra later, and the EU gap shows that AI features now launch region by region. If you sell into homes or build voice experiences, expect assistants that take actions in apps to become normal fast.
Watch Apple’s HomeView demo (1 min)
TRENDING AI TOOLS
Six new launches and updates you can try today. Click a name to open the tool.
🎭 Drama 3: Fish Audio's new text-to-speech model that you direct in plain words, like [barely holding it together], with sighs, pauses and whispers on single words. In preview on the free tier, which includes monthly credits.
🦙 Llama: A free 1 MB app from the llama.cpp team and Hugging Face that runs open models offline on your Mac or Windows PC and picks the right settings for your machine. Chat in the browser, no account needed.
🎬 Takes for Mac: Record talking-head videos with a teleprompter that follows your voice, then let Claude Code cut silences, add captions and draft the posts. Free app; needs Apple silicon and a paid Claude plan.
🎙️ OpenWhispr for iOS: A voice keyboard, notepad and meeting recorder for iPhone and iPad, with unlimited on-device transcription and no account. Free; Pro adds cloud sync and more meeting hours.
🎥 ngram 2.0: Turn a doc, URL or screen recording into a narrated explainer, product walkthrough or avatar video, then fix scenes by chatting. The free plan covers about 60 seconds of premium footage.
🧒 Gakku: An AI writing and speaking coach for kids aged 6 to 16 that asks questions instead of doing the work for them. Launched this week, with a 3-day free trial before $29.90 a month.
QUICK HITS
🛒 Shopify lets any AI agent shop its catalog: Shop MCP connects agents to more than 1 billion products, carts and Shop Pay checkout in one link. The shopper still confirms every purchase.
💳 Amex wants agents to prove what you asked for: Its new Intent Intelligence Framework records your instructions, such as brand, size and budget, and can stop a checkout that doesn't match them.
🎵 Deezer will let you hide AI-generated music: A toggle arriving before the end of 2026 removes fully AI-generated tracks, albums and artists from search.
That's it for today!
See you soon,
Olle Hellman | Founder of The AI Field