AI 24 Weekly Recap — 14 Nov–21 Nov 2026
Top of the week: GPT-6.2 Astra and Sol, Gemini 4 Neon, open-weight Tencent Hy4, Genie 4 in the Gemini API
Version 1
This week
01OpenAI launches GPT-6.2 Astra and GPT-6.2 Sol
GPT-6.2 Astra (frontier) and GPT-6.2 Sol (efficient, everyday work) arrived in ChatGPT Work, Codex and the API. Astra API price: $8 input / $40 output per 1M tokens, 20% below GPT-6 Astra.
Original update and sources →02Google DeepMind releases Gemini 4 Neon
Gemini 4 Neon sits between Gemini 4 Carbon and Hydrogen. It costs $3 input / $15 output per 1M tokens with a 1M-token context window, rolling out in the Gemini API, AI Studio and the Gemini app.
Original update and sources →03Tencent releases open weights for Hy4, its flagship multimodal MoE
Hy4 has 770B total and 49B active parameters with a 1M-token context window. Weights are on Hugging Face, code on GitHub, with a Hy4 System Card.
Original update and sources →04Google opens Genie 4 world model in the Gemini API and AI Studio
Genie 4 builds a world from a text prompt or a reference image and streams steerable 24 fps video with stereo audio. It requires a paid-tier Gemini API key.
Original update and sources →05Thinking Machines releases Inkling 2 Mini with open weights
Inkling 2 Mini is a 26B-parameter looped transformer designed to run on a laptop. Weights are on Hugging Face under Apache 2.0.
Original update and sources →06Ideogram releases Ideogram 4.6 image model
Ideogram 4.6 generates images in native 4K with no upscaling. Edits show less pixel drift outside the edited area and better color and material accuracy.
Original update and sources →07Kling 4.0 adds native 4K and 10-bit HDR video output
Kuaishou's video model Kling 4.0 now generates video in native 4K. 10-bit HDR output is available in both 4K and 1080p.
Original update and sources →08OpenAI brings GPT-Live-1.1 voice model to the API
GPT-Live-1.1 is a full-duplex voice model with improved pronunciation, accent consistency and transcription. A new standby option pauses inference after 5–60 seconds of idle time.
Original update and sources →09OpenAI API adds discounted 2h and 8h background windows
In Responses API background mode for GPT-6.2 Astra and Sol, a 2h completion window gives 30% off Standard token rates and an 8h window gives 40% off. Public beta.
Original update and sources →10Anthropic brings Claude computer use to Team and Enterprise in beta
Computer use lets Claude see a user's screen and work in the desktop apps the user allows. It is in beta on Claude Team and Enterprise plans.
Original update and sources →