Key Highlights
- V4-Pro model receives 75% price reduction valid through May 5, 2026
- API cache hit pricing reduced by 90% across the entire platform
- Two model variants available: Pro edition and Flash edition
- Built for compatibility with Huawei chip architecture while achieving top performance in open-source world-knowledge tests
- Pricing strategy reflects escalating competition within the global AI industry
Hangzhou-based AI developer DeepSeek has announced a substantial 75% price reduction for its recently unveiled V4-Pro model, reflecting the accelerating competition across the artificial intelligence sector.
The company introduced this promotional pricing structure for developers during the previous week. This limited-time offer extends through May 5, 2026, concluding at 15:59 UTC.
The revised pricing structure brings input costs for cache misses down to $0.435 from the previous $1.74. Cache hit charges decrease to $0.03625 from $0.145, while output fees fall to $0.87 from $3.48.
DeepSeek simultaneously introduced a 90% reduction for input cache hit charges throughout its complete API portfolio. According to the company, this adjustment became effective immediately and delivers substantial savings for users submitting recurring or similar queries.
The V4-Pro model arrives after considerable anticipation. Engineers optimized it for Huawei chip infrastructure, addressing circumstances where US export controls have constrained Chinese firms’ access to American semiconductor technology.
Dual Model Approach
DeepSeek offers the V4 series in two distinct configurations. The Pro configuration delivers enhanced capabilities and carried higher pricing before the discount period began. The Flash configuration provides a streamlined, budget-friendly alternative.
According to DeepSeek, the Pro configuration surpasses competing open-source models in world-knowledge evaluation metrics. Only Google’s proprietary Gemini-Pro-3.1 achieves superior results in these assessments.
The company positions the V4 models as optimized for AI agent applications. These frameworks manage sophisticated operations beyond basic conversational interfaces, though they demand additional computational resources.
This pricing initiative follows the debut of DeepSeek’s R1 model, which sparked widespread cost-based competition throughout the AI sector upon its release during the prior year.
Industry-Wide Pricing Dynamics
Numerous AI enterprises are transitioning from experimental phases to practical deployment of large language models. Reducing inference and operational expenses has emerged as a critical competitive strategy.
DeepSeek’s pricing adjustments are anticipated to prompt similar moves from competitors, particularly within China, where companies are developing alternatives to Western technologies.
American technology export restrictions have influenced this transformation, accelerating the growth of domestic AI infrastructure throughout China.
OpenAI, Anthropic, and Google continue launching new models at a rapid pace. Accessing these platforms often involves substantial costs, positioning DeepSeek’s reduced pricing as a compelling alternative.
The 75% reduction for V4-Pro continues through May 5. The comprehensive API price adjustments spanning DeepSeek’s full model range are currently in effect.
