BLAST RADIUS

Open-weight & owned stack · Quantization efficiency

User demonstrates Qwen Coder cluster inference

Sep 28, 2026

A user reported running quantized Qwen3-Coder-Next across five BC-250 boards at roughly 40 tokens per second with a 30,000-token context.

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

More in Open-weight & owned stack