News for hackers. Stuff that matters.
Re:Qwen/Qwen3.8-2.4T-A95B by simonw on Wednesday August 12, 2026 @11:38AM (#49274149)
Is this the largest ever open weight model release by parameter count? I think it is.
Re:Qwen/Qwen3.8-2.4T-A95B by NitpickLawyer on Wednesday August 12, 2026 @11:40AM (#49274181)
No, Kimi k3 is 2.8T params. This is 2.4T params but ~5TB weights because it was released in bf16 and ~2.5TB for the fp8 version. Kimi k3 launched with QAT 4bit, so ~1.5TB weights.
Re:Qwen/Qwen3.8-2.4T-A95B by Mercuriusdream on Wednesday August 12, 2026 @12:53PM (#49275364)
KIMI K3 was the biggest open weight release afaik; It is 2.8T-A100B if I'm correct