It seems that part of GPT 5.6’s token efficiency came from reducing the required “thinking budget” for high/xhigh. So, xhigh is no longer truly xhigh.

I’d recommend trying OpenAI models. The flexibility with your own harness is worth it. 5.3 Codex was the first time they had a well-polished coding model, imo. But it always needed high/xhigh. 5.4 was much better on a lot of fronts: - Better usability in terms of personality. - Much better technical capability than 5.3 on hard tasks - Can do a lot on high/medium, xhigh not needed and 5.4 mini can replace 5.3 Codex easily. 5.5 fixed a lot of the “early stopping”, and had better depth and technical capability. The main advancement was token efficiency and medium/low(!) working for 90% of tasks. 5.6 is much more in depth with tasks, works well on low reasoning levels, and seems to have significantly better capabilities. View quoted note →

5.6 feels much better than 5.5. It goes into further depth on a task, and has very little “early stopping”.

Did damus relay die?

@jb55 Could I get a dump of the Damus relay, along with the last dump from the last nuke?

Good morning ☕️

gm nostriches

Unfortunate to see, but oh well. If you want to save your notes, sign up to Nostr.land to import your notes with a click and 0 hassle. It is the longest operating paid relay, having done so for over 3 years.

nostr.land

Subscribe to Nostr.land

View quoted note →

Why is the developer of CLINK trying to sabotage Bolt12 by spreading FUD about LND introducing onion relay?

The American mind cannot comprehend this

👀

gm 👀 👀

@OceanSlim You can use ncrypt-go in Grain instead of a custom implementation for NIP-44.

Redirecting...

Many ecash projects are solutions in search of nonexistent problems

gm nostriches 👀