8 ms·
They are comparing it to 2 and 3 version old flash/fast versions of models but purely for tok/s. Then only comparing it to Mercury 2 on intelligence. This is
by ltbarcly3 8d ago
They are comparing it to 2 and 3 version old flash/fast versions of models but purely for tok/s. Then only comparing it to Mercury 2 on intelligence. This is very misleading and I suspect this model is basically useless.
- swiftcoder 8d agoEven pretty dumb models are useful for running subtasks (especially at this sort of speed). Note that in the coding section they only mention using it as a subagent for a smarter model