BLAST RADIUS

Open-weight & owned stack · Training tuning

UkisAI Releases Faster Swift-Qwen3.8-27B Model

Sep 14, 2026

UkisAI released Swift-Qwen3.8-27B, claiming nearly twice the inference speed with substantially fewer reasoning tokens and minimal accuracy loss.

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

Also covering this

More in Open-weight & owned stack