AI Digest: NVIDIA CEO Rumors, Gemini, Kimi, MiniMax, and OpenAI Updates
AI Digest: NVIDIA CEO Rumors, Gemini, Kimi, MiniMax, and OpenAI Updates
Key AI developments from November 2–15: rumors of a Jensen Huang leadership shake-up, new Gemini and Kimi models, MiniMax image generation, and OpenAI's Codex retro console.
AI Digest: November 2–15
Drama
- Rumors circulated that Jensen Huang stepped down as NVIDIA CEO after the board launched an investigation into a scandal involving infidelity, relationships with subordinates, and espionage; some claims were reportedly denied. Against this backdrop, the company lost more than a trillion dollars in market capitalization. While the board searches for a replacement, CFO Colette Kress is leading the company. There are rumors that Huang is trying to return.
LLMs
- Inkling 2 Small — Thinking Machines released weights for a 594B-parameter MoE model with 17B active parameters. The NVFP4 version requires two B300 GPUs or four H200 GPUs.
- Gemini Decision 1 — Google now has its own Jev.
- Gemini 4 Carbon — now available in the paid API and in Gemini for Ultra subscribers. Until November 30, pricing is $2/$10 per million input/output tokens, then $4/$20.
- Fable 5.2 — generates answers up to 30% faster than 5.1. In the API, pricing is $8/$40 per million input/output tokens.
- Kimi K3.1 Pro — Moonshot released the model weights and replaced Gated MLA with Gated DSA for more efficient long-context handling.
Generative Models
- MiniMax H3-Image — open weights, native 4096×4096 generation in six steps, and support for up to 16 references.
- FLUX 3 Max — fal offered a free trial of video generation with sound for up to 15 seconds in 720p from text or images.
- Gemini Omni 1.2 Flash — generates and edits video with sound for up to 15 seconds. In the API, pricing is about $0.07 per second in 720p, with input tokens billed separately.
Other
- Codex Retro Console — OpenAI released ModRetro Chromatic for $200. Users can build their own games in Codex and load them onto a cartridge.
Why it matters
AnalysisThe digest highlights how fast the AI sector moves, from leadership rumors at a major chipmaker to new open weights, faster models, and lower-cost generative tools.
Discuss in community
Share your questions and insights with developers
Related in Ecosystem
Practical guides, core terminology, and agent skills
LLM (Large Language Model)
A fundamental class of neural network architectures based on autoregressive transformers, predicting the probabilistic distribution of subsequent tokens and demonstrating emergent properties of abstract reasoning, code synthesis, and logical inference.
Jev and System One AI: A Complete Practical Guide to the Decision Model from TypeSafe
A deep engineering breakdown of the flagship Jev model: System One architecture versus LLMs, Choice/Score/Noul primitives, request parallelism, confidence mathematics, video use cases, code, and an overview of local open-weight alternatives (Laya, GLiNER2.5, CLM-8B).
Part I. AI Agent Architecture: The Foundations
The eleven foundational building blocks of modern autonomous agents: model taxonomy, the agentic runtime loop, advanced prompting, distillation, skills architecture, memory systems, function calling, and FinOps.
Frontier Models
The most powerful class of artificial intelligence at the forefront of global research (Claude 3.7 Sonnet, OpenAI o3/GPT-4.5, Gemini 2.0 Pro), defining the limits of modern reasoning, autonomy, and coding capabilities.