NVIDIA Qwen3.8-Flash-Next Achieves Over 16,000 Tokens/Second Throughput on GB300 NVL72

NVIDIA has announced that Alibaba's latest preview model, Qwen3.8-Flash-Next, is now supported on the NVIDIA GB300 NVL72 platform. The model has a total parameter scale of 176 billion, with approximately 6 billion parameters activated per token. It natively supports a context of 262,000 tokens and can be extended to 1 million tokens via YaRN, primarily targeting long-context agent applications such as intelligent programming, document processing, and tool invocation. NVIDIA stated that Qwen3.8-Flash-Next employs a mixed architecture of Gated DeltaNet (GDN) and Qwen Sparse Attention (QSA) to reduce computational and KV cache overhead in long-context scenarios. Testing shows that on the GB300 NVL72, the model achieves a single GPU throughput of over 16,000 tokens per second, with single-user throughput exceeding 200 tokens per second; it also supports inference frameworks such as SGLang, vLLM, and TensorRT-LLM.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.