TLDR
- DeepSeek has introduced a 75% price reduction on its V4-Pro AI model, valid through May 5, 2026
- API input cache hit pricing has been reduced by 90% across the company’s entire platform
- The V4-Pro is available in two configurations: a full Pro edition and a streamlined Flash edition
- Built to run on Huawei chip infrastructure, the model achieves top performance among open-source alternatives in global knowledge tests
- These aggressive price reductions represent the latest development in an accelerating AI pricing battle between Chinese and Western technology firms
Hangzhou-based AI developer DeepSeek has announced a dramatic 75% price reduction for its recently unveiled V4-Pro model, marking another significant move in the ongoing competitive battle between Chinese and Western artificial intelligence companies.
The promotional pricing for developers went live last week and will remain in effect until 15:59 UTC on May 5, 2026.
With the new pricing structure, input costs for cache misses have fallen from $1.74 to $0.435. Cache hit pricing dropped from $0.145 to $0.03625, while output costs decreased from $3.48 to $0.87.
Beyond the V4-Pro discount, DeepSeek has implemented a 90% reduction in input cache hit rates across all API offerings. According to the company, this change is already active and will benefit users who frequently submit similar or recurring requests.
The V4-Pro model arrives after an extended development period. Notably, it has been optimized for Huawei’s chip technology—a critical consideration as US export controls have restricted Chinese firms’ access to advanced American semiconductor products.
Dual Configuration Strategy
DeepSeek’s V4 lineup features two distinct configurations. The Pro variant delivers enhanced capabilities and carried premium pricing before the discount implementation. The Flash variant offers a more compact, budget-friendly alternative.
According to DeepSeek’s internal benchmarks, the Pro configuration surpasses all competing open-source models in world-knowledge evaluations. Only Google’s proprietary Gemini-Pro-3.1 achieves higher scores in these assessments.
The company positions its V4 models as purpose-built for AI agent applications. Such systems handle significantly more sophisticated operations compared to conventional chatbot interfaces, though they demand substantially greater computational resources.
This pricing initiative follows the introduction of DeepSeek’s R1 model, which sparked widespread price competition throughout the AI sector upon its release last year.
Industry-Wide Pricing Pressure
As the AI industry transitions from experimental phases to commercial deployment of large language models, reducing inference and operational expenses has emerged as a critical competitive strategy.
DeepSeek’s aggressive pricing is likely to force competitors—particularly those operating in China—to implement similar reductions as they develop alternatives to Western-dominated technologies.
American technology export restrictions have accelerated this dynamic, catalyzing rapid growth of independent AI development ecosystems throughout China.
Companies like OpenAI, Anthropic, and Google continue releasing advanced models at a rapid pace. However, premium pricing for these platforms creates strategic opportunities for competitively priced alternatives like DeepSeek.
The 75% promotional discount for V4-Pro continues through May 5. The comprehensive API price reductions affecting DeepSeek’s complete model portfolio are currently in effect.
