Skip to main content
AI RADARModels
352•08/11/2026, 17:15•4 min read

Neurodigest Weekly (#126)

ModelsGotBurnout Radar

Neurodigest Weekly (#126)

AI Daily Digest•Verified Tech Release
Executive TL;DR30s read

A roundup of key AI and technology news from the past week, including new models and updates from leading companies.

Neurodigest Weekly (#126)

LLM

  • Muse Code and Spark 1.2 — Meta has introduced its agent and a new model. Thanks to the ability to train on user data, the price has been reduced to $0.1/$0.2 per million tokens.
  • Qwen 3.8-Max — Alibaba has released a multimodal model with 2.4T parameters (95B active). It excels at visual agentic tasks and autonomous coding, priced at $2/$6. They promise to release the weights later this week.
  • First SSI Model — Rumor has it that Ilya Sutskever's startup plans to release its first model in August, without waiting for the announced superintelligence.

Generative Models

  • Lift4D — A framework for 4D reconstruction of dynamic objects and their textures from regular video footage.
  • Grok Video 1.5 — xAI has released a minor update for its video generator.
  • FLUX 3 Video — The video model from BFL is now available in API (up to 20 seconds with sound and lip-sync). A draft mode has been introduced, with a base price starting at $0.17 per second at 720p.
  • SeedRealtime — ByteDance has introduced a native full-duplex model. The assistant continuously watches the video stream, listens, and can initiate dialogue without delays.
  • Wan 3.0 — Alibaba has launched a public beta. The model generates videos up to 30 seconds in 1080p, allowing for the mixing of text, photos, videos, and audio for accurate transfer of appearance and style.
  • Wan Animate 2 — An alternative to pose-control for motion transfer based on video references without skeletons. Weights and nodes for ComfyUI have been released.
  • Grok Imagine Image 2.0 — xAI has significantly enhanced its image generator: added Magic Wand, precise area editing, background removal, and multi-reference up to 5 images.

Other

  • Gemini Robotics 2 — Google DeepMind has updated its VLA model: it now controls the entire humanoid body (including fine motor skills), coordinates a swarm of robots, and has an on-device version for quick adaptation.
  • Shifts at Google — Demis Hassabis has stepped down as CEO of DeepMind (transitioning to Chair and Chief Scientist of Alphabet), and Jeff Dean has left the company after 27 years to start Discovery Loop.

Why it matters

Analysis

This digest contains crucial information about new technologies and models in the AI field that could impact the future development of the industry. Keeping track of such news helps stay informed about the latest trends and innovations.

Discuss in community

Share your questions and insights with developers

Join Telegram
+5 Points
← Back to news feed

Related in Ecosystem

Practical guides, core terminology, and agent skills