CNBeta — 2026-04-27#
Top Story#
According to a cnbeta report, DeepSeek has aggressively slashed its API pricing for the V4 model series by up to 90%, bringing the cache-hit cost down to an astonishing 0.02 RMB per million tokens. This dramatic price war, fueled by architectural optimizations and deep integration with Huawei’s Ascend hardware ecosystem, aims to lower the barrier for agentic applications and significantly undercuts global competitors, signaling a major push by Chinese models to dominate the developer market.