DEEPSEEK V4 MODELS UNDERCUT COMPETITION ON PRICING

DeepSeek's latest models establish new price floors across their performance tiers. The V4 Flash variant, positioned as the budget option, costs significantly less than competing budget models from other providers. At $0.14 per million input tokens, it delivers substantial savings for high-volume token processing. The V4 Pro model targets users requiring higher capability levels, priced at $1.74 per input million and $3.48 per output million. This positions it as the cheapest option in its performance class, according to analysis by Simon Willison. The pricing structure reflects a widening gap in AI model costs. DeepSeek's aggressive rates contrast with established providers like OpenAI and Anthropic, which charge notably higher per-token rates across their model lineups. This pricing strategy extends DeepSeek's push into Western markets, where cost-conscious developers and enterprises evaluate model selection. Output tokens cost double the input rate across both models, a standard structure in the AI API market that accounts for increased computational requirements during generation. The release signals continued competition around AI model pricing and accessibility. As more models enter the market, per-token costs have trended downward, though substantial gaps remain between premium and budget tiers. DeepSeek's positioning suggests the company prioritizes market penetration through pricing rather than premium positioning. For developers evaluating models, the pricing advantage comes alongside considerations of model capability, latency, and reliability. Both V4 variants represent options for builders weighing cost against performance requirements across different use cases.