Hacker News new | past | comments | ask | show | jobs | submit
simonw | on: GLM-5.1: Towards Long-Horizon Tasks
Not only did this one draw me an excellent pelican... it also animated it! https://simonwillison.net/2026/Apr/7/glm-51/
stingraycharles | parent | next
Surely at this point it’s part of the training set and the benchmark has lost its value?
ipsum2 | parent | next
It made it realistic. A pelican is much more likely to be flying in the sky than riding a bicycle.
_pdp_ | parent
Simon, you need to come up with improved benchmarks soon.
lemonish97 | root | parent | next
Agree. But you can keep the pelican theme in whatever new benchmark you choose to come up with. Iconic at this point.
fancy_pantser | root | parent
let me see Tayne with a hat wobble