OK, fine, Qwen 3.8 Flash Next is nice
3.8 27b at Q6 kv8 was my driver before and that works super well, but Flash Next in strata IQ4 xs is really winning me over.
Proof1 community thread
Key points
- I held off as long as I could, but finally gave in and set up strata.
- Its not perfect, but it feels like models I was using professionally last year, which were/are good enough to do just about anything.
- You all are a bad influence.
Sources (1)
- [1]OK, fine, Qwen 3.8 Flash Next is nicer/LocalLLaMA (top, daily) · Oct 21, 03:53 PM
3.8 27b at Q6 kv8 was my driver before and that works super well, but Flash Next in strata IQ4 xs is really winning me over.
I held off as long as I could, but finally gave in and set up strata.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 20, 2026Voxtral 3
- Oct 20, 2026[AINews] DeepSeek DeepInfer: Open Source SOTA Inference running 155T toks/day/cluster making $22m/day at ~80% margin for a 81 day payback
- Oct 19, 2026MiniMax M3.1 Flash shows China's AI price race has no floor in sight
- Oct 17, 2026Alignment on SOTA Chinese Models like Qwen 3.8 Flash Next / GLM 5.3 Flash
- Oct 17, 2026Strata - Qwen 3.8 Flash Next - boosting decoding speed Tip
- Oct 17, 2026Nemotron 3.5 Bolt, Muse Spark Open, and a Fast LFM2.5 24B
