Is the AI Price Slayer Stopping? DeepSeek Rarely Announces 'Significant Price Hike' for API — Foreign Media Reveals the Reasons Behind

China's AI startup DeepSeek is changing its once-low-cost API pricing strategy. On the 6th, DeepSeek announced on its official API pricing page that it plans to raise prices across all API services in the near future. The company explicitly warned users that "the expected increase will be substantial" and advised them to "plan your usage accordingly." However, the exact percentage of the increase and the effective date have not yet been disclosed. For enterprises and developers heavily relying on DeepSeek's API, this means they must re-evaluate their future usage costs.

According to DeepSeek’s announcement: "We plan to generally increase the pricing of DeepSeek API services in the near term, with a substantial expected increase. Please plan your usage appropriately." Final details will be confirmed in an official notice.

How cheap is DeepSeek right now?

Based on DeepSeek’s current published rates, the V4-Flash model costs 0.02 CNY per million tokens for cached input hits, 1 CNY for misses, and 2 CNY for output. The V4-Pro model is priced at 0.025 CNY, 3 CNY, and 6 CNY respectively. Both models support a 1-million-token context window, with a maximum single output of up to 384,000 tokens.

Additionally, according to Singapore’s Lianhe Zaobao, DeepSeek currently uses a peak/off-peak pricing model. During peak hours—9 AM to 12 PM and 2 PM to 6 PM on weekdays—prices are double the off-peak rate. This means that if base prices rise across the board, high-volume developers could face even higher actual costs.

Why is DeepSeek raising prices now?

DeepSeek recently launched the public beta of its V4-Flash production API and stated in its official update log that the V4-Pro production version will be "released as soon as possible." Industry observers in China interpret this broad price adjustment as possibly linked to the imminent V4-Pro launch and a reallocation of computational resources. However, DeepSeek has not confirmed a direct connection between the price hike and the new model.

Notably, DeepSeek has long attracted attention for its low API prices. Founder Liang Wenfeng previously explained the pricing philosophy: "Our principle is not to operate at a loss, nor to seek excessive profits. This price is just slightly above cost with minimal margin." The company’s rare announcement of a significant price adjustment raises questions about whether DeepSeek will maintain its low-price advantage under the new pricing structure.

Will DeepSeek still be cheap after the hike?

According to Reuters, citing tests by research firm Artificial Analysis, the average cost per benchmark run for V4-Flash is approximately $0.03 USD—less than 1% of the ~$3.15 cost for Anthropic’s Claude Fable 5. However, overall capability scores still lag behind top-tier models from OpenAI and Anthropic.

More exclusive reports from Feng Media: · OpenAI Just Cut Prices! DeepSeek Immediately Counters with '4x Cheaper Than Cabbage' Pricing—Foreign Media: 'Price War Is Over' · Did OpenAI Launch a 'Self-Attack' on a Major AI Open-Source Community? U.S. Think Tank Coalition Sends Letter to Trump, Demanding Federal Investigation · Beijing Watch: Did DeepSeek’s Leaked Audio Become Evidence for U.S. Sanctions? Liang Wenfeng Claims '4 Huawei Cards Beat 1 Nvidia,' While ChangXin Surges 466% Post-IPO for Emergency Rescue!