还是那个问题:电脑性能开销 还有 token开销 如何

[louishino]

[JerryZLiu]

If you're running gemini, it's very efficient, probably only a few % off your battery for an entire day. Takes about 1 million input tokens and 100k output tokens per hour.

[limberc]

I am using Qwen3-VL 30B with my Macbook Pro.

My work mainly involves doing some research and local testing for deep learning research and development. This requires me to open a large number of Chrome tabs as well as multiple PyCharm/VScode projects. In addition, I need to test the status of my other models (I need to manually offload the models to load other large models). Overall, it does not cause any delay in my computer's response.

Most of the analysis is correct but when I write code, it will recongize my Python code as React or JS.