Home | Notifications | New Note | Local | Federated | Search | Logout

Note Detail

lain@lain@fediffusion.art (2026-10-07 19:17:42)
inteersting comparison between opus 5.5 and qwen 3.8 flash next on the same task: https://deepu.tech/local-ai-qwen3.8-flash-next-best-local-llm/
---Reply---
underproofed ganbold@ganbold (2026-10-07 19:24:52)
seeing the stats especially the time to PR, I think qwen 3.8 flash next is good for like a "leisure PR" where you just queue an issue then it will fetch and work on a single issue at a time. opus is more like, immediate or something need to be patched right now vibes.

are you running the local inference with laptop, lain?

---Replies---

lain@lain@fediffusion.art (2026-10-07 19:26:04)
@ganbold yes, and my experience is very similar to the one in that blogpost, although i usually don't use qwen for coding because i have the claude and openai subscriptions. i do use it for agentic work though, where i don't really care if it takes 30 seconds or 10 minutes.