Community · Oct 29, 2026, 21:40 UTC

Swift 1.5 Flash Next Heretic v2 quant released for Qwen 3.8 with reverse engineering focus

A community GGUF quantization of a Qwen 3.8 based model, with reworked calibration and bit allocation tuned for coding and reverse engineering. Quantized with a heretic v2 weight set from another Hugging Face user, and may need a Strata BF16 kernel workaround on load.

• LabAlibaba
• ModelQwen 3.8 Swift 1.5 Flash
• Topicrelease
• Topicopen-weights
• Topiccoding