DeepSeek(@deepseek_ai):💾 Smaller KV cache. Bigger savings. Compared with the previous generation, V4.1-Flash’s KV cache needs just: 🔹 1/4 the HBM 🔹 1/8 the SSD storage Cache-hit charges often account for a large share of agent costs. Compressing the cache cuts those costs significantly. 3/6

DeepSeek

@deepseek_ai

Unravel the mystery of AGI with curiosity. Answer the essential question with long-termism.

参加 October 2023

0 フォロー中 1.1M ファン

DeepSeek@deepseek_ai

2026.09.10 06:10

💾 Smaller KV cache. Bigger savings. Compared with the previous generation, V4.1-Flash’s KV cache needs just: 🔹 1/4 the HBM 🔹 1/8 the SSD storage Cache-hit charges often account for a large share of agent costs. Compressing the cache cuts those costs significantly. 3/6