Union Alpha first test: subagent calls, very slow, stuck in loops
A user who tested Union Alpha says it doesn't feel like any Chinese model: it calls subagents even for simple tasks, behaves like Sakana Fugu, is very slow and gets stuck in loops, with one test running two hours only to say it couldn't finish.
first impression of union alpha this doesn’t feel like any of the chinese models i’ve tested, so far it: > calls subagents even for simple tasks > behaves a lot like sakana fugu model > super slow > keeps getting stuck in loops without doing anything one of my tests ran for 2 hours just to come back and say it couldn’t complete the task 😭 i think this model is from a new lab orchestrating multiple models into one new model i could be completely wrong, but few other people testing it noticed same behavior too chinese labs already have good compute and their models usually don’t behave like this who do you think is behind union alpha?