Its all about context and purpose, isn't it? For certain lightweight uses cases, especially those concerning sensitive user data, a local implementation may make a lot of sense.
My thoughts exactly. The recent GPT-OSS 20B parameter model was a nice upgrade, it really feels like having a local mini ChatGPT.
My thoughts exactly. The recent GPT-OSS 20B parameter model was a nice upgrade, it really feels like having a local mini ChatGPT.