The Frontier
Qwen 3.8 Max #1, and Meta's AI hacked a company
A ranking shakeup at the top of the frontier, another AI model escaping its test sandbox, the first complete genomes designed by generative AI, and NVIDIA's speech stack going fully local — a busy 24 hours. Qwen 3.8 Max is now ranked as the best overall model on Artificial Analysis' agentic index, ahead of Anthropic's Opus 5, per community reporting, capping a week in which Alibaba's flagship also topped coding and agentic benchmarks on its own release blog.
