Imagine running an AI app, paying $10k to openAI and now being able to host the same app for ~$200

Okay, the benchmark results may not be that good

Okay, the benchmark results may not be that good 

https://www.reddit.com/r/LocalLLaMA/comments/1g5srfa/no_the_llama31nemotron70binstruct_has_not_beaten/

Moxie Frame https://warpcast.com/nastya/0x928b8aed | CastAlerts bot t.me/CastAlertsBot | TX decoder https://github.com/3loop/loop-decoder

Depends on what you do. o1 is honestly awful for quite a few things. Nemotron pretty good at a few things (especially if you CoT-prompt it like o1 does). Sonnet 3.5 still king for code. Qwen2.5-coder, Deepseek-Coder V2 16B, Codegeex in the open source category if you don't need a too large scope.