Updated 3 November 2026. This month's AI news moved fast, even by AI standards. Google and OpenAI released new models, Google started showing ads inside its AI search answers in India, the Indian government went looking for more AI computing power, and Anthropic filed to go public. Here is what happened, and what it means if you run a business.
The short version
- Google now shows ads inside AI Mode answers in India. It was announced on 28 October, and existing Search, Shopping and Performance Max campaigns are eligible automatically.
- Gemini 4 Argon, Google's newest top model, became generally available on 13 October. Its introductory API price ends on 30 November.
- OpenAI released GPT-6.1 Terra on 26 October. It is a mid-sized model built to read screens and operate software, at half the price of the model it replaces.
- DeepSeek published an open-source coding agent that watches developer chats and offers fixes before anyone asks. For now it is a research prototype.
- Anthropic filed to list on Nasdaq and reported $16.3 billion in revenue for the first half of 2026.
- The IndiaAI Mission is buying more compute. C-DAC has invited providers to supply up to 10,000 more GPUs and similar chips, with all data kept in India.
1. Ads arrive inside Google's AI Mode in India
On 28 October Google said it is scaling ads in AI Mode to ten more countries: Australia, Canada, India, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore. AI Mode is the chat-style answer view in Google Search. Brazil, Mexico and the United Kingdom are due before the end of the year. The ads are labelled "Sponsored". In the US, Google is also adding ads to follow-up questions, so a sponsored result can appear at each step of a conversation, not only beside the first answer (Google).
Google says AI Mode queries are on average three times longer than traditional searches. People describe their whole problem, such as "a CRM for a 20-person sales team that works with WhatsApp", instead of typing two keywords.
What to do: According to Google's help page, existing Search, Shopping and Performance Max campaigns are eligible automatically, so there is no new campaign to set up. In India these ads run in English (Google Ads Help). Watch your search terms report over the coming weeks for longer, conversational queries. Make sure your landing pages answer the whole question, not just the keyword. For organic traffic, pages that answer specific questions plainly are easier for AI answers to quote and link to.
2. Gemini 4 Argon is now available to everyone
Gemini 4 Argon, Google's newest frontier model, became generally available on 13 October. Developers and companies can use it through the Gemini API, AI Studio and Google Cloud. Google AI Pro and Ultra subscribers get it in the Gemini app and in AI Mode in Search. Google aims it at long, complex work such as coding, finance, legal and cybersecurity tasks. A single request can hold up to one million tokens of input and output combined, which is room for hundreds of pages of documents (Google).
The price is worth noting. Until 30 November it costs $2 per million input tokens and $10 per million output tokens through the API. After that it doubles to $4 and $20. If you are testing it for a project, this month is the cheap time to do it. For how Google's assistant compares with OpenAI's for everyday business use, see our Gemini vs ChatGPT comparison.
3. OpenAI's GPT-6.1 Terra: computer use at a lower price
On 26 October OpenAI released GPT-6.1 Terra. OpenAI's models come in four sizes, Luna, Terra, Sol and Astra, from smallest to largest, and Terra is the second. OpenAI describes it as a fast model for visual reasoning and computer use. That means it can read what is on a screen and take actions across apps: filling in forms, reading technical drawings or drafting proposals. It is available in ChatGPT Work, Codex and the API, at half the price of GPT-5.6 Terra, which it replaces (9to5Mac).
Why it matters: a model that can operate ordinary software can automate jobs that have no API. One example is copying details from emailed purchase orders into an old ERP screen. For high-volume tasks like that, a cheaper mid-sized model usually makes more sense than the most powerful one.
4. DeepSeek's open-source "Super-agent"
On 31 October the Chinese AI lab DeepSeek published a prototype coding agent. Instead of waiting for a question, it watches developer group chats, spots coding problems and proposes fixes. It is open source under the MIT licence, but DeepSeek calls it research software, not a finished product (Bloomberg, Digital Trends).
The wider trend is proactive AI that acts on what it sees instead of only answering prompts. That is useful, but it means deciding in advance what an agent may read and what it may change. DeepSeek itself says the agent runs with full access to the computer's user account and should be isolated before use (Dataconomy). That is good advice for any AI agent: give it its own account, with only the permissions the job needs.
5. Anthropic files for an IPO on Nasdaq
Anthropic, the company behind the Claude models, filed publicly for an initial public offering on 26 October and plans to list on Nasdaq under the ticker ANT. Its filing shows $16.3 billion in revenue for the first six months of 2026, against $4.6 billion for the whole of 2025 (Yahoo Finance). An amended filing put third-quarter revenue at about $17 billion, 47% more than the quarter before (Reuters via The Globe and Mail).
For other businesses the point is simple. Companies are now paying for AI at scale, not just trying it out, and the tools your competitors use are improving every few weeks.
6. IndiaAI Mission seeks up to 10,000 more GPUs
Under the IndiaAI Mission, C-DAC has invited AI service providers to supply up to 10,000 more "AI compute units": GPUs, TPUs and similar accelerators for ministries, researchers, startups and industry. All infrastructure, data, logs and backups must stay in India. IT Minister Ashwini Vaishnaw said the first tender would be for 5,000 of them (MediaNama).
This adds to the more than 38,000 GPUs already onboarded on the mission's common compute facility. They are offered to Indian startups and academia at an affordable rate (Digital India).
If you are an Indian startup training or fine-tuning your own models, the subsidised compute is worth a look. For everyone else, the requirement that data stays in India shows where the rules are heading, so know where your AI provider processes your customers' data.
What this means for your business
- AI keeps getting cheaper. Terra costs half of what its predecessor did, and Gemini 4 Argon starts at $2 per million input tokens. If a support chatbot or document-processing tool looked too expensive before, price it again. Work out the cost per conversation or per document before you decide.
- Agents that operate software are ready for small pilots. Start with one repetitive task, have a person approve what the agent does, and measure the time saved.
- Search is turning into a conversation. With answers and ads inside AI Mode, your website needs pages that answer real questions clearly. Your Google Ads account needs watching for longer, conversational searches.
- Don't tie yourself to one model. The top model changes every few weeks. Build your AI features so that the model behind them can be swapped without rewriting the application.
- Know where your data goes. Check which provider processes your customers' data, in which country, and whether it is used for training.
Frequently asked questions
Does Google show ads in AI Mode in India?
Yes. On 28 October 2026 Google said it is scaling ads in AI Mode to India and nine other countries. The ads are labelled "Sponsored", and existing Search, Shopping and Performance Max campaigns are eligible automatically.
How much does Gemini 4 Argon cost?
Through the API it costs $2 per million input tokens and $10 per million output tokens until 30 November 2026. After that the price rises to $4 and $20.
What is GPT-6.1 Terra?
It is OpenAI's mid-sized model, released on 26 October 2026 and built for visual reasoning and computer use. It is available in ChatGPT Work, Codex and the API at half the price of GPT-5.6 Terra.
Is DeepSeek's Super-agent ready for business use?
Not yet. DeepSeek describes it as research software. It runs with full access to the computer's user account, so DeepSeek says it should be isolated before use.
When did Anthropic file for its IPO?
On 26 October 2026. It plans to list on Nasdaq under the ticker ANT, and its filing reported $16.3 billion in revenue for the first half of 2026.
How can Indian startups get affordable GPUs?
The IndiaAI Mission's common compute facility has more than 38,000 GPUs onboarded. It offers them to Indian startups and academia at an affordable rate.
How Bitinfusion can help
We build AI chatbots that answer from your own content and AI automation that cuts manual back-office work. We also handle SEO and Google Ads for businesses adapting to AI search. If you want to know which of this month's changes are worth acting on for your business, book a free call.
Prices are in US dollars, as published by each provider on the dates above. AI products change quickly, so check the linked sources for the latest details.