@imnotchalk Zai are underrated. GLM5.3 and Flash are wonderful for more than just coding, I use them for internet searches and general lore research and assistance.
Deepseek models are great at coding, math, general reasoning tasks, but hallucination rates are too high for users like me.
@DonJoewonSong Playground Games have lost passion, and need a restructure. The reliance on temp employees has harmed them, and lack of clear direction too. Support also seems to be non-existent, from an already poor showing, it's somehow even worse now. I still can't see treasure chests.
@bindureddy Will High reasoning stay locked behind a paywall?
Astra and many other models have broken toggles for search, playground, code execution. I think it's mainly models from OpenAI.
Glm 5.3 flash is missing vision.
We really need ways to disable models as-well, pushing 50 total.