AI benchmarks are getting WILD.
🤯 GPT-5.6 Sol: 91.9% on Terminal-Bench Claude Opus 5: still at the frontier GLM-5.3-Flash: insanely cheap + punching WAY above its weight MiniMax M3: 114 tokens/sec ⚡ So instead of posting another boring benchmark table... I animated the
Today's news is one that changes the level of expectations about what an AI is capable of doing.
Claude Opus 5 observed an animation in p5.js, understood the mathematics behind the points and in seconds generated a version of itself, but with an octopus.
Fable 5.1 and Opus 5.1 need to beat Grok 4.7 on cost per task or Anthropic’s reputation is cooked.
Fable 5.1 is rumoured to ship at the same $10 / $50 per million tokens as Fable 5. If that holds, they did not fix the thing anyone was actually complaining about. Nobody said
Worldfold Engine updates over the last weeks for my zamta_io project!
I build a complete PBR ggx material system with SPOM displacement, a sky occlusion SSGI and reflection systems. It's getting harder to hold all together just with vibecoding ( Opus and Fable 5). But it works.
You have reached the end of the archive
All of Claude Opus 5