Comment by bigyabai

2 months ago

There are already local models that surpass 4.6 in benchmarks, they're just extremely difficult to run.