pull down to refresh

1 sat \ 2 replies \ @optimism 2 Feb \ parent \ on: Step 3.5 Flash: Fast Enough to Think. Reliable Enough to Act. AI

Performs as good as GLM 4.7 for what?

write
preview
1 sat \ 1 reply \ @cd300cc371 OP 2 Feb

Coding seems to the primary use case, atleast according to benchmarks. (which can be gamed, unfortunately)

reply
1 sat \ 0 replies \ @optimism 2 Feb

I haven't had a single code battle in arena.ai this week where GLM-4.7 won

reply