Newsletter Digest — June 15, 2026

This week’s AI landscape was dominated by breakthroughs in open‑weight models, extreme‑scale efficient inference, and new visions for AI ecosystems and safety.

This week’s AI landscape was dominated by breakthroughs in open‑weight models, extreme‑scale efficient inference, and new visions for AI ecosystems and safety.

Frontier Open Models

  • GLM‑5.2 emerges as the leading open‑weight model, outperforming GPT‑5.5 on knowledge‑work benchmarks and ranking top in coding arenas, with a 744B‑parameter MoE, 40B active parameters per token, a 1M‑token context window and IndexShare sparse attention 12.
  • Z.ai’s trajectory suggests an open Fable‑class model could appear by December, alongside recent releases such as Laguna M.1 and North Mini Code 1.

AI Infrastructure & Efficiency

  • Xiaomi unveiled a 1‑trillion‑parameter model that runs at 1000 tokens/second using FP4 quantization and speculative decoding, pushing the limits of efficient large‑scale inference 3.
  • Anjney Midha’s AMP proposal describes a cross‑cloud system operator that pools compute to raise GPU utilization from <10% to 60‑70%, aiming to extract more useful FLOPs from existing hardware 4.

Applications & Research

  • Radical AI’s self‑driving lab produced ~1,200 new alloys in six months, nearly ten times the pace of prior efforts, and open‑sourced tools such as TorchSim and the MATRIX dataset 5.
  • Midjourney previewed a prototype full‑body ultrasonic CT scanner (the Midjourney Scanner) targeting low‑cost, radiation‑free imaging with eventual AI‑assisted analysis, though speed, resolution and regulatory hurdles remain 6.

Policy & Ecosystem

  • Satya Nadella’s Loopcraft article argues that the real AI opportunity lies in building frontier ecosystems and learning loops that combine human and token capital, highlighting the Anthropic Fable export‑control crisis and advances in inference efficiency 7.
  • Researchers launched the nonprofit Sequent to develop alignment techniques for superintelligent AI, seeking $100‑150 M to pursue theory bets such as scalable oversight and heuristic arguments 3.

Sources

  1. Latent Space — [AINews] GLM > GPT? GLM-5.2 passes vibe check; Z.ai forecasts Open Fable by December
  2. Latent Space — [AINews] GLM-5.2: the top Frontend Coding model in the world, IndexShare for Speculative Decoding
  3. Import AI — Import AI 461: "Alignment is not on track"; FrontierCode; and synthetic research interns
  4. Latent Space — The Professor of Outputmaxxing — Anjney Midha, AMP
  5. Latent Space — 🔬 The Self-Driving Lab — Joseph Krause, Radical AI
  6. Latent Space — [AINews] Midjourney Medical: scan your organs like you step on a scale
  7. Latent Space — [AINews] Satya on Loopcraft: Building Frontier Ecosystems