China's biggest open-weight models landed overnight, Anthropic's IPO numbers are now public, and Indian states are writing their own AI rulebooks before the Centre finalises one.
Two massive open-weight models from China just dropped: Hy4 (770B) and Kimi K3.1 Pro (2.8T)
Tencent released Hy4, a 770 billion-parameter mixture-of-experts model with a 1 million-token context window and built-in visual understanding and computer-operation skills. Hours later, Moonshot AI pushed Kimi K3.1 Pro — a 2.8 trillion-parameter model — to Hugging Face, also with a 1M-token context window. Both are free for anyone to download and run.
Why it matters: Open-weight models this size used to be a US-only story. When frontier models are freely available, startups and researchers everywhere — including India — can build on them without licensing fees or API costs. The gap between proprietary and open AI is narrowing fast.
ChatGPT can now make animated GIFs. A new coding model just got 3x faster at a fifth of the price.
OpenAI updated ChatGPT Images to version 2.6, adding animated GIF creation across ChatGPT and its Codex developer tool. Separately, Cognition released SWE-2 Flash, a 276 billion-parameter coding model that runs three times faster and costs about five times less than its flagship version. These are the kinds of incremental updates that quietly make AI tools more useful for everyday work.
Why it matters: GIF generation sounds small, but it closes the gap between what a graphic designer and a text-only AI tool can do. Cheaper, faster coding models mean AI-assisted software development becomes accessible to smaller teams and solo developers.
Anthropic's IPO numbers are out: $17B in revenue, $23B in losses, and a $895B compute bill
Anthropic filed an amended S-1 with the SEC on November 2, adding third-quarter results. Revenue was roughly $16.96 billion — about 12 times higher than the year before. The company also posted a $23.3 billion net loss, mostly from non-cash accounting charges tied to convertible notes. The filing shows Anthropic has committed to nearly $895 billion in computing costs, with about $70 billion due in the next year. It plans to list on Nasdaq under the ticker "ANT."
Why it matters: Anthropic is growing at a pace few companies in history have matched, but its spending on compute is on a completely different scale. This IPO, if it goes through, will be one of the largest tech listings ever. It also shows just how capital-intensive building frontier AI really is — this is not a software business with normal margins.
Bihar and Goa both launched state AI policies this week — before the national framework is finalised
Bihar's Cabinet approved its AI Policy 2026 on November 2, offering financial incentives including SGST reimbursements for AI companies, startups, and Global Capability Centres (GCCs) setting up in the state. Two days later, Goa announced its own AI Policy 2026 with a ₹250 crore innovation fund and a target to train 50,000 AI-ready professionals by 2030. Both states are moving ahead without waiting for MeitY's national consultation paper, which was promised within a month of its October 8 announcement.
Why it matters: State-level AI policies create immediate, local opportunities — especially for GCCs and startups in those states. If you are building an AI company in India right now, understanding what your state government is offering is becoming as important as following the central government.
The EU's AI labeling rules are now live. Every AI-generated image, video and text needs a disclosure marker.
Article 50 of the EU AI Act came into force on August 2 and has been in a grace period ever since. That grace period for machine-readable markers on existing AI systems runs out on December 2. Any company deploying AI tools to European users — including Indian software firms serving European clients — needs to have disclosure and transparency mechanisms in place before that deadline.
Why it matters: EU rules tend to become global standards because companies build one compliant system and roll it everywhere. If you work at a company that uses AI outputs in customer-facing products for European clients, compliance deadlines are now close enough to matter.
Anthropic's IPO roadshow is coming — and it could set the valuation for all of AI
Anthropic has filed its S-1, the numbers are out, and a roadshow — where the company pitches institutional investors before listing — is the next step. Reports suggest it could list as early as this month at a valuation between $1.5 and $2 trillion. That figure would make it one of the most valuable companies ever to go public. How investors price it will signal what the market thinks frontier AI is worth, and that will ripple through valuations across the whole industry — including smaller AI startups in India and elsewhere.
IndiaAI Fellowships close Nov 30 • 125,000+ jobs tracked as AI-displaced this year
The IndiaAI Fellowship 2026–27 offers up to ₹2 lakh for postgraduate students in AI-related fields — applications close November 30. Apply if you are eligible. Meanwhile, the Alliance for Secure AI says it has tracked more than 125,000 jobs attributed to AI displacement as of June 2026, and that number keeps rising.