Google released Gemini 4 Hydrogen on 12 November, a Gemini 4 model Google says is built for speed and efficiency in everyday coding and agent work. It is priced at $1.50 per million input tokens and $7.50 per million output tokens, flat across its 1 million-token context window.
Google releases Gemini 4 Hydrogen
Google's announcement, by Koray Kavukcuoglu, says Hydrogen is built for "rapid execution" while Gemini 4 Carbon and Argon handle the hardest reasoning. Google says it scores 51 on the Artificial Analysis Intelligence Index at its default high thinking level, against 50 for Argon, and that it finishes tasks 2.5 times faster than Gemini 3.8 Flash at that default setting. It is available across the Gemini API, Google AI Studio, Gemini for business, the Gemini app (free users get it when auto model selection picks it) and Google Search, with four thinking levels (low, medium, high and a new xhigh).
Why it matters: if you run a chatbot or automation on Gemini 3.8 Flash, Hydrogen is the model Google is pointing you to. Test it on your own tasks before trusting the benchmark numbers.
Gemini 2.5 API calls get rerouted from 1 December
Google's Firebase AI Logic model page says that from 1 December 2026, Gemini Developer API requests to general-use Gemini 2.5 Pro will be routed to gemini-3.8-flash, and 2.5 Flash to gemini-3.5-flash-lite. Google warns this "may cause breaking changes." On the Agent Platform Gemini API (formerly Vertex AI), the page says all general-use 2.5 models have already shut down.
Why it matters: anything still calling Gemini 2.5 will change behaviour in about two weeks without anyone touching it. Find those calls and test the newer model now.
OpenAI starts processing API requests inside India
OpenAI said on 12 November that data sent through its API in India will now be processed in the country, with inference for supported workloads also running there, Business Standard reports. It covers GPT-6 Luna at first, through the Responses and Chat Completions APIs, and eligible customers turn it on by creating a new API project with India as the region for storage and processing and using the India regional endpoint, Analytics India Magazine reports. Nitin Bawankule, OpenAI's head of enterprise for India, said some Indian businesses "have not fully deployed AI for important work" because their data has to be processed in the country.
Why it matters: if data rules kept you to pilots, ask whether GPT-6 Luna covers your use case. Only that one model is included at launch.
ChatGPT Wallet lets OpenAI's agents check out, with your approval
OpenAI launched Wallet on 13 November (US time) for a limited number of US ChatGPT Pro subscribers aged 18 and up with a US-issued Visa, Mastercard or American Express card, TechCrunch reports. Its agents, called dots, can help pick items and complete the purchase. Users review the items, total and card, then approve with Face ID or a fingerprint, and the card is charged through a payment token, so the card details aren't shared with the dot or the merchant. OpenAI plans to add spending limits and rules for pre-approving purchases later.
Why it matters: if you sell online, OpenAI has published a Commerce Partner Playbook for sites that want to support shoppers using agents. It's US-only for now, but worth a read.
Microsoft says its email outage came from a network upgrade
Microsoft apologised for the Outlook outage that ran across two workdays and said it "was a technical issue not related to cyber or AI / agents," ABC News reports. Microsoft said the incident "was introduced by an issue during an earlier network capacity upgrade that surfaced following a recent network change," and that it has addressed the underlying issue. As of that 13 November (US time) statement, it was still restoring service to remaining customers and regions.
Why it matters: Microsoft traced this one to its own network change. Whatever the cause, plan how your team works for a day without its main provider.
Fewer Indian firms say they are fully ready for cyberattacks
Kyndryl's third annual Readiness Report, which surveyed 3,700 senior leaders in 21 countries including 200 in India, found 35% of Indian organisations said they were completely ready to manage cyberattacks, down 10 points from last year, CXOToday reports. The figure for organisations globally was 26%, down 11 points. Kyndryl sells IT infrastructure and security services.

Why it matters: AI tools add new ways in. Before you connect one to email, files or payments, decide who checks its access and how you'd switch it off.
Worth watching: the 1 December Gemini 2.5 cutover, and whether OpenAI adds more models to India processing.
Arthat AI
Working on something similar at your company?
Arthat AI builds these automations for Indian businesses every week. If you want to talk through whether this fits your situation, we'll give you a straight answer — even if the answer is "don't build it."


