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

Note Detail

Reply to @harblinger@wizard.casa
lain@lain@fediffusion.art (2026-09-23 06:05:14)
@harblinger seems like everyone is doing 'cheaper and better' now, trying to reduce inference and training cost. probably a good idea, given the quality of frontier models.
---Reply---
Harblinger@harblinger@wizard.casa (2026-09-23 06:06:39)
flash ALL the models!
---Attachments---
image: https://wizard.casa/media/a2acb59f49bf29f0aab460710df4685ea1c8ce8ccfcdd36ecbed4ad55e22148d.png

---Replies---

lain@lain@fediffusion.art (2026-09-23 06:08:10)
@harblinger still amazed how good the qwen flash is. i use it for tons of stuff and it all runs on my own systems.