Companies are racing to incentivize employees to use AI. But as some companies are finding, the more employees that use the technology, the heavier the bill.
I mean eventually open weight models will be good enough to do the few things AI is actually good at, and then there’s another reason to not pay astronomical prices.
I don’t know shit about fuck when it comes to servers and networking, so my use case is a self hosting assistant, help me understand proxmox, write yaml, maintain documentation, configure .arr stack etc.
Oh wait. I have started using it for something. I use it to look over contracts as a first pass and ask it to point out any red flags. I still read the contracts, but it can help find obvious unacceptable terms.
I switched las month from GPT-OSS-120B to Gemma4 31B. For some simple scripts I found the latter considerably mor efficient, less verbose and with better results. At the same time the sycophancy is much worse.
No way I’ll use cloud based models and feed the data base. But also local models clearly like to burn energy.
But the benchmarks are semi useless. Performance at new problems in your precise use case matter and that’s only something one can find out you using it. This can differ wildly from what benchmarks suggest. AI can be an ability enhancer when for example someone who has no idea hos to code suddenly manages to create crappy code but that manages to do a job where for example success can be easily verified. But if the result should ve good or even reliable … well…
Price per intelligence benchmark has been going down though but openai killed 4o mini which was the absolute cheapest smartish model.
The writing is not on the wall just yet, but with some players dropping out and some enshittification I could see models becoming very expensive.
I mean eventually open weight models will be good enough to do the few things AI is actually good at, and then there’s another reason to not pay astronomical prices.
What’s your use case? To me, lazy api doc search, and grammar/flow check
I don’t know shit about fuck when it comes to servers and networking, so my use case is a self hosting assistant, help me understand proxmox, write yaml, maintain documentation, configure .arr stack etc.
I don’t use AI for anything.
Oh wait. I have started using it for something. I use it to look over contracts as a first pass and ask it to point out any red flags. I still read the contracts, but it can help find obvious unacceptable terms.
Good one. Any sort of “give me a rough idea of…” works. But I don’t trust it for anything slightly serious
I switched las month from GPT-OSS-120B to Gemma4 31B. For some simple scripts I found the latter considerably mor efficient, less verbose and with better results. At the same time the sycophancy is much worse.
No way I’ll use cloud based models and feed the data base. But also local models clearly like to burn energy.
But the benchmarks are semi useless. Performance at new problems in your precise use case matter and that’s only something one can find out you using it. This can differ wildly from what benchmarks suggest. AI can be an ability enhancer when for example someone who has no idea hos to code suddenly manages to create crappy code but that manages to do a job where for example success can be easily verified. But if the result should ve good or even reliable … well…