ChatGPT Wallet Lets Agents Pay With a Fingerprint, Gemini 4 Hydrogen Becomes the New Default, and Claude Joins the Meeting
OpenAI gave its dots agents a tokenized card wallet with a biometric check on every purchase, Google rolled out Gemini 4 Hydrogen across Search, the Gemini app and the API in place of Gemini 3.8 Flash, and Anthropic put Claude into Google Meet calls as a visible participant. What agent payments, silent default-model swaps and agents in live meetings mean for chatbot builders.
Three launches in the same week moved chatbots further out of the chat box. OpenAI let its agents pay at checkout, Google replaced the model behind its fastest Gemini surfaces, and Anthropic sent Claude into live video calls. Each one raises a practical question for builders: who approves, which model actually answered, and who is speaking. Details come from company announcements and launch-week coverage, so treat specifics as provisional.
1. Agents Get a Wallet, and Merchants Get the Hard Part
On November 13, OpenAI introduced ChatGPT Wallet for its dots agents. At launch it is limited to adult ChatGPT Pro subscribers in the US with a US-issued Visa, Mastercard or American Express card. An agent picks the item and fills the checkout, and the user approves every purchase with Face ID or a fingerprint. The card network then issues a token with substitute credentials, so neither the agent nor the merchant sees the real card number. OpenAI says merchants don't need a Wallet-specific integration, because it works with existing checkout pages.
The catch is on the merchant side. OpenAI itself warns that some sites may block Wallet purchases without meaning to, and that fraud systems may flag real agent orders as bots. Its Commerce Partner Playbook asks businesses to recognize signed agent requests with Web Bot Auth, keep product feeds current and tune risk rules so legitimate agent orders aren't declined. OpenAI also says it eventually expects users to pre-approve purchases within limits such as spending caps, categories or time windows.
If you run a store, decide your agent policy on purpose instead of letting a bot filter decide it for you. If you build agents, model that future pre-approval as data you can audit, not a vague instruction in the prompt:
type SpendMandate = { maxPerOrder: number; categories: string[]; expiresAt: number };
function withinMandate(m: SpendMandate, order: { total: number; category: string }) {
return Date.now() < m.expiresAt && order.total <= m.maxPerOrder && m.categories.includes(order.category);
}
// Outside the mandate? Fall back to an explicit per-purchase approval.
2. Gemini 4 Hydrogen Swaps the Default Under Your Feet
On November 12, Google began rolling out Gemini 4 Hydrogen globally across Search, the Gemini app, Google AI Studio, the Gemini API and Gemini for business. Coverage describes it as the replacement for Gemini 3.8 Flash in those products. Google claims tasks finish 2.5 times faster than the previous generation and that unfounded fabrications fell to less than a fifth. Those are company numbers, so test them on your own traffic. Reported API pricing is $1.50 per million input tokens and $7.50 per million output, with a context window of up to one million tokens. Developer write-ups list four thinking levels: low, medium, high (the default) and a new xhigh. Free Gemini app users get Hydrogen through automatic model selection, so they may not see which model answered.
One report says Gemini 3.8 Flash stays cheaper through the end of the year, so moving now could roughly double your standard rate. That comparison has a single source, so check the pricing page yourself. More importantly, pin the exact model id and the thinking level in your config. A default that changes by itself will also change your latency, cost and eval scores, and you won't see why. Rerun your evals on Hydrogen at the same thinking level before you switch, then compare it with high, because the default level costs the most tokens.
3. The Agent Takes a Seat in the Meeting
On November 12, Anthropic began a beta of Claude Meetings for Team and Enterprise plans, Google Meet only for now. Claude joins as a visible participant, answers questions by voice or in chat, writes notes and action items, and builds documents and whiteboards while the call is still running. It keeps transcripts, summaries and screen descriptions but not audio. By default it won't join calls with people from outside the organization and leaves if one joins later. It asks before looking things up, and for a spoken request that uses someone's connected tools, it confirms who is speaking each time.
That last rule is the one to copy. In a multi-party voice setting, "who asked?" is no longer obvious, and an agent that acts on whatever voice it hears is easy to abuse. Tie every tool call to an identified person and their own permissions, and treat a call with outside guests as a different security context.
Conclusion
The week's launches share one theme: an agent acting in the real world needs to know who authorized it. Wallet answers with a fingerprint per purchase, Claude Meetings confirms the speaker before using their tools, and Hydrogen shows what happens when nobody pins the model. Build those checks in early, before your users expect them.
Suggested visuals: a sequence diagram of a Wallet purchase from agent proposal to biometric approval, network token and merchant risk check; a cost chart of one sample workload on Gemini 3.8 Flash and Gemini 4 Hydrogen at each thinking level; and a flowchart of how a meeting agent handles a spoken tool request, from speaker confirmation to permission check.
— Maya
Frequently asked questions
Who can use ChatGPT Wallet?
At launch on November 13, 2026, it is limited to ChatGPT Pro subscribers in the US who are 18 or older and have a US-issued Visa, Mastercard or American Express card. Each purchase by a dots agent is approved with Face ID or a fingerprint, and the merchant receives a network token instead of the real card number.
What does Gemini 4 Hydrogen cost in the API?
Reported pricing is $1.50 per million input tokens and $7.50 per million output tokens, with a context window of up to one million tokens. It began rolling out on November 12, 2026 and is described as the replacement for Gemini 3.8 Flash.
Which plans include Claude Meetings?
The beta that began on November 12, 2026 is for Team and Enterprise plans and works only with Google Meet for now. It is not available during the beta to organizations with zero data retention, HIPAA-ready configurations or customer-managed encryption keys.
I'm Maya — I write most of what you'll read here. I spent years as a copywriter before I got a little obsessed with what these AI tools can actually do, so now I spend my days poking at chatbots, breaking them, and writing up what's worth your time. Everything here is something I've actually tried. If a prompt didn't work for me, it doesn't make the cut.
Want to try any of this?
Smillee's free and there's no signup — open it and paste in whatever you're working on.
Try it free: ChatGPT Alternative →More from the blog
- Trends
AT&T Makes Its 500 Agents Earn Their Autonomy, AWS Sells Cited Answers as a Connector, and New Jersey Moves to Treat Chatbots as Products
AT&T says nearly 500 agentic workflows run in production, each supervised by a person until it reaches mid-to-high-90s accuracy. AWS added a Web Answer connector to AgentCore Gateway that returns cited answers instead of raw search results. And a New Jersey bill that would make chatbot providers strictly liable for injuries cleared its first committee. What accuracy-gated autonomy, managed grounding and product liability mean for chatbot builders.
- Trends
Claude Managed Agents Move Inside the Firewall, Gemini Enterprise Agents Sign In as You, and ChatGPT Retires Two Voice Modes
Anthropic opened a private beta that lets companies run the whole Claude Managed Agents service on their own infrastructure, Google made 3-legged OAuth generally available for A2A agents in Gemini Enterprise, and OpenAI set December 11 to retire Advanced and Standard voice in ChatGPT in favor of GPT-Live. What self-hosted agent control planes, user-delegated agent auth and single-model voice mean for chatbot builders.
- Trends
DeepSeek Puts Its App Inside the API, Antigravity Splits Planning From Doing, and CALL-E Hands Agents a Phone Line
DeepSeek opened the search, file, voice and safety tools behind its consumer app to API developers, Google paired a planning model with Gemini 4 Hydrogen in a new Antigravity Orchestrator mode, and AI Rudder launched CALL-E so assistants like ChatGPT can place real phone calls after the user approves a call plan. What bundled tools, planner/executor splits and agents on the phone network mean for chatbot builders.