# Amid AI Industry Chaos, Microsoft and OpenAI Drop New Models for Coding and Work

> Source: <https://www.cnet.com/tech/services-and-software/openai-chatgpt-astra-6-2-microsoft-ignite-news/>
> Published: 2026-11-18 16:28:38+00:00

**Key takeaway:** Microsoft's seven new MAI models, including MAI Code 2 and MAI Cyber 2 Flash, are pitched as significantly cheaper to run at enterprise scale than OpenAI's and Anthropic's leading models, and its licensed weight program lets companies customize them.

It has been a wild week for AI, and it’s only Wednesday. Aside from [outages](https://www.cnet.com/news-live/apple-outage-11-17-2026/), [cheating scandals](https://www.cnet.com/tech/services-and-software/jensen-huang-big-tech-learn-from-nvidias-mistakes/) and [upcoming IPOs](https://www.cnet.com/tech/services-and-software/anthropic-ipo-filing-ai-bubble-hype/), there are several new models available now. Microsoft dropped seven new versions of its MAI models at its Ignite conference, and OpenAI released GPT-6.2 Astra and Sol, the next step forward for its flagship models.

First, Microsoft: At its Ignite developer conference on Tuesday, Microsoft [introduced](https://microsoft.ai/news/microsoft-ignite-2026/) seven new AI models. They include MAI Cyber 2 Flash, MAI Code 2, MAI Image 2.7 and Dragon Copilot Transcribe — models meant to help with tasks in [cybersecurity](https://www.cnet.com/tech/services-and-software/ai-agents-are-a-cybersecurity-nightmare-thats-only-just-begun/), coding, [image generation](https://www.cnet.com/tech/services-and-software/best-ai-image-generators/) and transcription. 

It’s clear from Microsoft’s marketing that it is aiming to position its new tech as a cheaper alternative to OpenAI’s and Anthropic’s leading models, with many of these significantly cheaper to run at the enterprise level. That, plus its new [licensed weight program](https://www.zdnet.com/innovation/microsoft-licensed-weights-ai-models-ignite-2026/), which lets companies customize the new AI models for their purposes, highlights how Microsoft is continuing to pitch its AI tech to the workplace.

More from CNET

Next, OpenAI: The ChatGPT maker dropped two new model variants of its flagship model, GPT-6.2 Astra and GPT-6.2 Sol. Astra is the big model — it costs the most and uses the most firepower to handle complex tasks — and Sol is the “everyday” version. Both models beat their sibling models in OpenAI’s benchmarking. These are available to paying subscribers in ChatGPT and Codex globally now. (Disclosure: Ziff Davis, CNET’s parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

## The AI elephant in the room

While these new models might be useful for AI power users, they are far from most of the AI industry’s mind this week. During Tuesday’s earnings call, new interim Nvidia CEO Colette Kress tried to assure investors that everything was “business as usual” after [weeks of turmoil](https://www.cnet.com/tech/services-and-software/nvidia-earnings-call-ai-economy-stress-test/) inside the company rocked the global economy. Nvidia cofounder and now-former CEO Jensen Huang’s multiple extramarital affairs, including one with [a woman suspected of ties to Chinese intelligence](https://www.cnet.com/tech/computing/nvidia-ai-spy-saga-jensen-huang-china-chips/), threw the company into chaos and have attracted the attention of [the Securities and Exchange Commission](https://www.wsj.com/business/sec-is-probing-nvidia-amid-departure-of-ceo-jensen-huang-ee7643f8) and both chambers of Congress, where Huang has agreed to testify.

And let’s not forget about Anthropic, the Claude maker plans to become a public company this week, with the possible biggest-ever IPO slated for Thursday. Anthropic’s reportedly seeking an opening price of $78 per share, for a total valuation of $1.4 trillion — narrowly beating the other largest-ever IPO earlier this year, [Elon’s Musk mega-conglomerate, SpaceX](https://www.cnet.com/news-live/spacex-ipo-live-starlink-xai-going-public-2/) (including now SpaceXSI, because [President Trump said so](https://www.cnet.com/tech/services-and-software/trump-ai-super-intelligence-internet-response/)).

The Nvidia scandal caused [a drastic 14% dip](https://www.cbc.ca/news/business/nvidia-jensen-huang-ceo-selloff-investments-retirement-9.7420093) in its stock price during its first week and wiped out $1 trillion in the company’s value, but Anthropic investors don’t seem to be worried. At least, not publicly.

If nothing else, Anthropic’s IPO and the new models from OpenAI and Microsoft are stark reminders that the tech industry is nowhere near ready to slow down on AI. Even as financial, [environmental](https://www.cnet.com/tech/services-and-software/data-centers-ai-climate-change-hyperscale/) and legal concerns mount — and arguably [hit a fever pitch](https://www.cnet.com/tech/services-and-software/midterm-election-2026-democrats-rein-in-big-tech-reaction/) during the midterm elections — Big Tech is plowing ahead with generative AI.
