Matt Wolfe's weekly AI news video spans a confirmed NVIDIA leadership change, two frontier-model launches, Adobe's agent-heavy MAX announcements and a long list of smaller product updates. The useful question is not whether this was a dramatic week. It is which facts are settled, which products you can use, and which claims still need qualification.
NVIDIA confirmed that Jensen Huang resigned and Colette Kress became interim CEO; a separate board inquiry remains an investigation. Anthropic released Fable 5.2, Google opened access to Gemini 4 Carbon and launched Hydrogen, and Adobe announced new creative agents with mixed release statuses. OpenAI's Decisions API segment in the video is sponsored. The practical rule: read the original launch page before treating a demo, preview or reported allegation as a finished fact.
Watch Matt Wolfe's AI News Roundup
NVIDIA: What Is Confirmed, and What Is Not
The hard news is in NVIDIA's November 12 release: Huang resigned as president, CEO and board member, effective immediately. CFO Colette Kress became interim CEO, and the board began a search for a permanent successor. The release did not give a cause for the resignation or say that the board asked him to leave.
Earlier, Huang's November 8 message acknowledged marital infidelity and said he would cooperate with the board inquiry. The board statement described an independent investigation following reports that he had been the target of a foreign-intelligence effort. It also said the board had no reason to believe national security or NVIDIA secrets were compromised. An investigation is not a finding of wrongdoing.
The video also discusses press reporting about personal relationships and China-bound chip shipments. The Information reports alleged links and shipment totals; those are its reported findings, not facts established by NVIDIA's announcement or a completed public inquiry. Wolfe's suggestion that the board pushed Huang out, and his forecast about a possible return, are interpretation and speculation. We should not infer the reason for a leadership decision from the timing alone.
What does change immediately is governance: a new interim leader and board chair have taken over while the company prepares to name a permanent CEO. NVIDIA says it expects customer commitments and its product roadmap to continue without interruption. Whether that holds, and whether the inquiry changes the picture, will require later evidence.
Fable 5.2, Gemini 4 Carbon and Hydrogen
Anthropic launched Fable 5.2 and Mythos 5.2 on November 9. Fable is available to users and developers for ambitious technical work; Mythos access is more restricted through verification programs. Anthropic says the models improve coding, reasoning and vision, and describes hardware and science demonstrations. Those demos are meaningful examples, but not a guarantee that a model will solve an arbitrary production task without review. Anthropic recommends starting with Opus 5.5 for most everyday coding and knowledge work.
Google made Gemini 4 Carbon available on November 10 after a narrower cybersecurity rollout. Developers can use the paid API tier; consumer access is tied to Google AI Ultra in the Gemini app and other Google products. The initial API rate was $2 per million input tokens and $10 per million output tokens, with Google saying it would rise to $4/$20 after November 30. Wolfe prefers Carbon's result in one of his own visual tests; that is a useful hands-on judgment, not a general benchmark win.
Hydrogen followed on November 12 as Google's faster model for everyday and agentic workflows. Its rollout spans developer, business, app and Search surfaces, but access differs by plan and product. For a real decision, run the same representative jobs through each model, record quality and cost, and check the result yourself. Leaderboard rank alone cannot tell you whether an 11-minute visual build or a day-long code task is the better fit.
A Smaller, Faster Decision Instead of a Long Answer
At 09:52, Wolfe moves into an OpenAI-sponsored segment about the Decisions API. Its three question types are predicate (whether a condition holds), choice (which option fits) and score (an ordered rubric). OpenAI documents a `gpt-6-luna` endpoint for bounded classification and routing; the low-latency and cost comparisons in the video are product claims, not an independent test in this article.
Google also launched its Gemini Decisions API this week, using `gemini-decision-1` for yes/no, choice and rubric-scoring questions. These endpoints can be useful when a workflow needs a structured judgment such as "route to support or billing" rather than an essay. For consequential decisions, define thresholds, log results, test calibration on your own data and keep a human fallback for ambiguous cases.
Adobe MAX and Claude Move Work Into the App
Adobe's MAX announcement expands its Creative Agent across Firefly and Creative Cloud, describes AI assistants in products including Premiere and Illustrator, and previews local MCP integrations for editable files. It also lists Firefly Design, Video and Graph workflows. Do not read the entire announcement as a generally available feature list: individual items are marked beta, preview or private beta. The practical question for a creative team is whether the agent can modify the native project without flattening layers, sequences or brand controls.
On the workspace side, Anthropic introduced Claude Whiteboards and Sheets in beta and made Dashboards generally available under the name Claude Data. Interactive connectors let partners place app experiences inside Claude. The same announcement calls Claude Meetings coming soon, despite the video's brief discussion of how it might work. These tools are worth testing on a small board, sheet or dashboard before replacing established team systems.
The Rest of the Week, Without the Hype
| Update | What the source says | Useful limit |
|---|---|---|
| Businesses in ChatGPT | OpenAI beta for claiming business information and seeing engagement insights. | Business Agents begin with eligible US ecommerce and online-service businesses; a claim is not a promise of traffic. |
| Muse Ember | Meta's smaller multimodal model for bounded, high-volume agent work. | Vendor examples and price-per-task claims need testing on your own workload. |
| Anthropic IPO roadshow | Proposed offering and expected $70-$78 price range were announced November 10. | The registration was not yet effective at announcement; a proposed range is not a final trading price. |
| MiniMax H3 Image | Model card provides weights and usage details for image generation and editing. | Local deployment requires substantial hardware; video speed claims are not measured here. |
| Automattic Spacefast | Hosting and publishing destination for work made with AI agents, with versions and access controls. | Spacefast says it is not itself an AI builder; review runtime and export needs. |
The most actionable pattern is clearer than any one headline: strong models are getting specialist companions for narrow decisions, creative suites are making agent edits more native, and distribution is becoming another surface for agents. Test each workflow at its actual access tier and keep results, costs and approvals visible.
Video Chapters
| Time | Topic |
|---|---|
| 00:00 | NVIDIA controversy and reported allegations |
| 06:32 | Jensen Huang CEO transition |
| 07:33 | What the transition could mean for AI |
| 09:52 | OpenAI Decisions API (sponsored segment) |
| 11:08 | Claude Fable 5.2 and Mythos 5.2 |
| 13:15 | Gemini 4 Carbon |
| 16:16 | Gemini 4 Hydrogen |
| 18:43 | Adobe MAX agents and creative tools |
| 21:36 | Businesses in ChatGPT |
| 22:15 | Meta Muse Ember |
| 22:33 | Gemini Decisions API |
| 22:51 | Anthropic IPO roadshow |
| 23:05 | Claude interactive connectors |
| 23:22 | Claude Whiteboards, Sheets and Data |
| 24:05 | MiniMax H3 Image |
| 24:42 | Automattic Spacefast |
| 25:17 | Final thoughts |
Original Sources and Editorial Limits
Checked 13 November 2026 against the supplied transcript and linked original announcements. The embedded video is Matt Wolfe's commentary; OpenAI sponsored its Decisions API segment. The NVIDIA personal-conduct and shipment details are reported allegations, and the board's investigation is ongoing. Model demos, vendor benchmarks and launch pricing may change; no independent benchmark or product trial was performed for this article.