NVIDIA Withdraws Qwen3.8-Flash-Next AI Model Profile Due to Stability Problems

AI-generated NewsSnap summary based on source reporting.
Published: 2026-09-01
Category: technology
Source: NVIDIA Developer Forums
Original source

NVIDIA has removed the Qwen3.8-Flash-Next ModelOpt NVFP4 TP2 experiment profile for two GB10 systems. This withdrawal follows the recurrence of an invalid-probability crash under specific multi-turn long-prefill traffic conditions, highlighting ongoing stability challenges in the deployment of advanced AI models.

Want more?

Open NewsSnap.ai for the full app experience, including audio, personalization, and more news tools.

Open NewsSnap.ai