Engro Holdings Reappoints Dawood Duo to Lead Next Chapter
May 27, 2026
Chinese AI start-up DeepSeek made its 75% V4‑Pro price cut permanent, slashing API costs to 0.025–6 yuan per million tokens, as reported by Reuters.
Chinese AI start-up DeepSeek has made its 75% promotional price cut on the flagship V4‑Pro model permanent, keeping fees at a quarter of their original level. The company announced the move in a statement on Saturday, as reported by Reuters.
DeepSeek reduced V4‑Pro API costs to between 0.025 and 6 yuan per million tokens (about $0.0035 to $0.83), down from 0.1 to 24 yuan previously. The start-up did not confirm whether improved supply of Huawei’s Ascend 950 chips enabled the permanent cut.
When DeepSeek launched V4 last month, it said the Pro version would cost up to 12 times more than the Flash variant due to “constraints in high-end compute capacity.” The company had also predicted Pro pricing would fall sharply once Huawei Ascend 950 supernodes launch at scale in the second half of 2026.
Third-party benchmark firm Artificial Analysis said the price cut places V4 Pro at the global frontier for cost-efficiency on an intelligence-per-dollar basis. DeepSeek’s model now costs about US$268 to run Artificial Analysis’ Intelligence Index benchmark tests.
OpenAI and Anthropic’s latest flagship models, GPT-5.5 and Claude Opus 4.7, cost about 12 and 19 times more respectively to perform the same benchmark task. Most other leading models on Artificial Analysis’ intelligence-per-dollar ranking are also from China, including MiniMax’s M2.7 and Xiaomi’s MiMo V2.5 Pro.
Huawei’s AI chip sales have benefited from U.S. export controls preventing Nvidia from selling advanced semiconductors in China. However, separate curbs on chipmaking equipment exports have limited Huawei’s ability to scale up Ascend production.
