DeepSeek V4 challenges ChatGPT with lower prices, stronger performance

ISLAMABAD: Chinese artificial intelligence company DeepSeek has launched its latest model, DeepSeek V4, with significantly lower pricing and competitive performance, posing a fresh challenge to premium AI tools such as OpenAI’s ChatGPT and other leading Western platforms.

The company released two versions of the model -- DeepSeek-V4-Pro and DeepSeek-V4-Flash -- on April 24, promoting them as high-performance systems for reasoning, coding, and AI agent applications while keeping costs far below top-tier rivals. 

According to DeepSeek’s official API pricing page, the company is offering a limited-time 75% discount on DeepSeek-V4-Pro until May 5, while also cutting input cache hit prices across its entire API suite to one-tenth of the original launch price.

As per reports, the move sharply increases pricing pressure in the global AI market, with DeepSeek specifically targeting developers building “agentic” AI applications, which require repeated prompts and high computing demand. The company said the lower cache pricing would make large-scale deployment significantly cheaper for enterprises.

DeepSeek says V4-Pro leads all current open models in world knowledge and trails only Google’s Gemini-Pro-3.1 among major models, while also outperforming current open-source rivals in math, STEM, and coding benchmarks. The company added that the model offers a default one-million-token context window, allowing users to process much larger documents and workflows than standard chatbot systems. 

Technology analysts and users on social media have also highlighted the aggressive pricing strategy.

Tech commentator Armaan Sidhu wrote on X that DeepSeek’s input cache pricing now costs only a fraction of fresh token rates, arguing that for applications with repeated prompts, the effective cost per query could fall by 60 to 90%.

He added that while premium players like OpenAI may still lead in frontier capabilities, “most production AI applications don’t need frontier capability. They need ‘good enough’ capability at scale.”

Another technology observer, Sébastien Muler, described DeepSeek V4 as a “1.6T parameters, open-source” model available at “1/10th the price of OpenAI,” saying small and medium enterprises can now access serious agentic AI at far lower cost.

Chinese users also praised the model’s writing quality and reasoning depth, particularly in Chinese-language creative writing and philosophical dialogue, with some users saying DeepSeek V4 had become stronger than Gemini and Claude for native Chinese content generation.

Reuters noted, however, that while DeepSeek V4 has received attention for pricing and hardware efficiency, the market response has been more subdued than the company’s earlier breakthroughs, with analysts saying competition among Chinese AI firms has intensified and efficiency gains are now expected across the industry.

Still, the release signals a broader shift in the AI race, where cost-efficient deployment may become as important as raw benchmark performance.

Industry observers say if DeepSeek can provide “good enough” performance at a fraction of the cost of premium U.S. models, businesses focused on customer support, document processing, software assistance, and automation may increasingly choose lower-cost open models over expensive frontier systems.