DeepSeek Cuts V4-Pro Prices 75%, Pressuring Google Cloud AI Competitiveness

GOOGGOOG

DeepSeek is offering a 75% discount on its V4-Pro AI model input tokens, reducing the cost to $0.036 per million through May 5, and has cut cache-hit rates across its API lineup to one-tenth of prior levels. These reductions intensify pricing pressure on Google’s AI services in the high-volume query segment.

1. DeepSeek Price Reductions

DeepSeek is offering a 75% discount on its V4-Pro AI model input tokens, reducing the cost to $0.036 per million through May 5, and has cut cache-hit rates across its API lineup to one-tenth of prior levels effective immediately.

2. Model Architecture and Performance

V4-Pro employs a mixture-of-experts architecture with 1.6 trillion parameters (49 billion active) and a one-million-token context window, while the lighter V4-Flash uses 284 billion parameters (13 billion active), positioning both as leading open-source alternatives.

3. Competitive Pressure on Google Cloud AI

These aggressive pricing moves target workloads with high volumes of repeated queries, directly challenging Google Cloud AI’s cost competitiveness and potentially forcing price adjustments or margin compression in the high-volume AI services market.

Sources

F