Termsslug: inference-compute

Inference Compute

推理算力

GPU resources and scheduling capacity backing online model serving, directly bounding concurrency and quotas.

Curated Intelligence

1 items · Page 1/1
IT之家 · IT业界T2·Industry
72 pts
DeepSeek Caps Long Chats and Daily Quota as Compute Strain Bites

DeepSeek has imposed usage caps on its app and web client, dropping long conversations inactive for over seven days, cutting single-input length to roughly 40,000 characters and limiting daily conversation quotas with a server-busy notice once exhausted. The move extends a stretch of compute strain: eight service outages since October, against weekly V4.1 Flash calls surging 801% to 33.6 trillion tokens on OpenRouter.