Skip to content

1.68.1 — Every model works: Google's limits read right

Latest
Latest

Choose a tag to compare

View all tags
@sppideey sppideey released this 06 Oct 11:18
v1.68.1
252f6d6

Every model except Flash-Lite looked "rate limited". Google wraps its errors in a list that ucode could not read, so a model's used-up daily allowance (about 20 a day on 3.8, 3.7, 3.6 and 3.5 Flash) was mistaken for a short per-minute wait and retried all day.

Now

  • Used up for today: ucode says so, says when the model comes back (12:30 PM in India), and offers to carry on with Flash-Lite, which has its own 500. /model marks it USED UP.
  • Pro on a free key: said to be not free, instead of retried forever.
  • Per-minute limits: waited out for exactly as long as Google says, and the next requests are spaced to fit.
  • Bad API key: says to check the key (ucode login). It used to call it a "malformed request" and squash the conversation.
  • Retired models: reported as retired, not retried.
  • Overloads: the newer Flash models and Gemma carry on with Flash-Lite while Google is overloaded.
  • Gemma: its thinking no longer shows up inside its answers.

Install or update: npm i -g ucode-agent@1.68.1

Assets 2
Loading