If you release it as open source, anyone can host it and publish the API.

So how do you make money? DeepSeek's answer is "because the operating costs are orders of magnitude lower." If you make something for a cost of 1 and sell it for a cost of 6, competitors that cost 20 won't be able to enter in the first place.

起業の履歴書Aug 164

AI Highlights of the Week: GLM-5.3 is launched on Zhipu, DeepSeek V4 Pro is heavily updated, and Google launches Gemini 3.7 Flash

Crypto KnightAug 161

BigMoeOnEdge: streaming MoE inference on Android, built on llama.cpp (ggml_org).

DeepSeek V4 Flash IQ2_M (92 GB) running on a 12 GB midrange phone at ~1 tok/s. Qwen3.6-35B-A3B up to 7-8 tok/s, Gemma 26B and Qwen3-30B in the same tok/s range. CPU only. Experts stream from UFS

Raffaele ManciniAug 162

Xianyu is the largest AI hotspot in China and the most profitable self-media platform in China🙃 The previous one is MiniMax H3

这次是DeepSeek Harness Ps: 现在还有人蹲Mac mini M4 吗？🙂

LonelyAug 163