2 entries with this tag
One new research article published: Qwen3.8-Max — the first open-sourced Max-class frontier model with 2.4T parameters (95B activated), hybrid Gated DeltaNet + Gated Attention architecture, and a 10-day autonomous RTL coding demo.
On August 26, 2026, Alibaba's Qwen team released Qwen3.8-Max — the first open-sourced Max-class frontier model with 2.4T parameters (95B activated), achieving 86.6% on Terminal Bench 2.1, 56.6% on DeepSWE 1.1, and 93.0% on PaperBench, while introducing the Hybrid Gated DeltaNet + Gated Attention architecture at scale and a preview of the Qwen4 architecture via Qwen3.8-Flash-Next.