AI Weekly: Cheap Tokens, Tight Safeguards, and a Two Million GPU Order
Week of August 26 to September 2, 2026 By Alex Merced, Data Lakehouse and AI Evangelist Three labs shipped models this week and every one of them led with price. Anthropic cut cache reads 75%, Z.ai put a natively multimodal 320B model on the table at $0.15 per million input tokens, and Alibaba previewed its next architecture with a model that computes six billion parameters per token. Underneath…
Three labs released new AI models in the week of August 26 to September 2, 2026, each emphasizing affordability. Anthropic reduced cache reads by 75%, making its Claude Fable 5.1 and Mythos 5.1 models accessible at $10 per million input tokens and $50 per million output tokens. Z.ai unveiled a multimodal 320B model called GLM-5.3-Flash on Hugging Face, offering native image and video input at a competitive price point.
Alibaba previewed its next-generation architecture with a model boasting six billion parameters per token. Concurrently, technology giants AWS and NVIDIA pledged to acquire an additional two million GPUs, while AMD introduced software version 10 tailored for agent-based systems. Meanwhile, MCP published a roadmap centering on agent identity for future specifications.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
