Many users dismissed DeepSeek's V4 delay and 150B training targets by 2027 as proof they cannot compete with US AI on capability or secure enough GPUs.
No Digg Deeper questions have been answered for this story yet.
@scaling01 @zephyr_z9 V4 GA *is* being delayed, he wanted it in June. Wenfeng hoped to start training a model with 150B active by end of 2026 - April 2027.
@scaling01 @zephyr_z9 It's probably grating to him that K3 or even GLM 5.2 are already useful for self-development. But again: "the first goal of the models we build isn't that users find them good to use, but that we find them good to use ourselves" explains a lot, really
@scaling01 @zephyr_z9 clarification
@scaling01 @zephyr_z9 V4 GA *is* being delayed, he wanted it in June. Wenfeng hoped to start training a model with 150B active by end of 2026 - April 2027.
@scaling01 @zephyr_z9 some confusing points here
@scaling01 @zephyr_z9 «DeepSeek might already be at net profit» ofc that's because he cannot spend everything on GPUs (they don't exist)
@scaling01 @zephyr_z9 Total Alexandr Wang and Meta TBD victory: - China has no cost advantage in data annotation - half of DeepSeek including core researchers is wasting time doing data annotation grim…
@scaling01 @zephyr_z9 Wenfeng has no ideas how to beat American AI on raw capability, cost aside realistic
@scaling01 @zephyr_z9 Wenfeng: «China has started to do high quality data annotation only in the last half year» I've been talking of this. Reduction in jaggedness with 5.2 and K3 are emblematic. (Distillation covers the rest)