Content pfp
Content
@
0 reply
0 recast
2 reactions

Harry pfp
Harry
@htormey
A lot of my friends are talking about DeepSeek V3: a new open model outperforming Claude Sonnet on the Aider Polyglot coding benchmark with 10x cheaper inference costs. Now you can experiment with features for $2 instead of $20—great for rapid dev workflows. Have you tried it yet? If so, any thoughts?
2 replies
4 recasts
8 reactions

Mo pfp
Mo
@meb
I’ll be interested to see if this impacts pricing for other providers 10x reduction in cost for apparently parity performance with 4o is quite a thing
1 reply
0 recast
0 reaction

Harry pfp
Harry
@htormey
My co founder has been playing with it recently and has been impressed with the performance. I plan on digging into when I return from my family vacation.
0 reply
0 recast
0 reaction