Trying out some visualisations for autoresearch to understand how models approach this.
Inferring and plotting references to previous trials is already interesting: both bigger and higher-effort models seem to make more longer-range inferences. Claude Opus 5 at low vs xhigh:
Grok 4.6 is here — and it's redefining cost-performance.
✅ 1.5T parameters (no parameter stacking) ✅ 1/4th the cost of Claude Opus 5 ✅ Extremely high visual ceilings ✅ Handles long-flow code like a pro
I created a 3D model of a fantasy world using Claude Opus 5.
It's not my own work, but I made it for verification purposes, so if you're interested, please take a look at the URL.
Trying to talk to claude Opus 5 to get work done be like 🤣🤣
You have reached the end of the archive
All of Claude Opus 5