Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

is there a primer somewhere on the security/privacy considerations of deepseek?

For the phone app does it send your prompts and information to China?

OpenRouter says if you use them that none of their providers send data to China - but what about other 3rd parties? https://x.com/OpenRouterAI/status/1883701716971028878

Is there a way to host it yourself on say a descent specd macbook pro like through HuggingFace https://huggingface.co/deepseek-ai/DeepSeek-R1 without any information leaving your computer?



Running it in a MacBook Pro entirely locally is possible via Ollama. Even running the full model (680B) is possible distributed across multiple M2 ultras, apparently: https://x.com/awnihannun/status/1881412271236346233


That’s a 3 bit quant. I don’t think there’s a theoretical reason you couldnt run it fp16, but it would be more than two M2 Ultras. 10 or 11 maybe!


Well there's the practical reason of the model natively being fp8 ;) One of the innovative ideas making it so much less compute-intensive, apparently.


the 70B distilled version that you can run locally is pretty underwhelming though


These Janus models are small enough to self-host locally.

You can also self-host a smaller distilled DeepSeek R1 variant locally.


This is not comparable to the R1 DeepSeek language model which has over 600 billion parameters. This for image generation and is 7b parameters and will run on most recent computers locally.


Good luck with that.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: