BLAST RADIUS

Open-weight & owned stack · Quantization efficiency

Ninfer user reports 600-token Qwen throughput

Sep 17, 2026

A user reported serving Qwen3.6 35B-A3B at approximately 600 tokens per second for one request with Ninfer on an RTX Pro 6000.

Read the original at reddit.comOpens the publisher's site in a new tab

More in Open-weight & owned stack