Dev2h ago
Detected, not submitted
Swift-Qwen3.8-27B
Own Swift-Qwen3.8-27B?
Verify your site to claim this listing and get emailed the moment a buyer's looking for something like it.
A post-trained open-source language model that reduces overthinking tokens by 58% while maintaining accuracy, delivering 1.95x faster inference speeds.
Spotted on Hacker News on September 16, 2026.
open-source
KI
kisjovan
Maker