DeepSeek V4 Pro leads cost-efficiency rankings after 75% price cut
Synopsis
DeepSeek has made a 75% price cut on its V4 Pro model permanent, making it 12x cheaper than OpenAI's GPT-5.5 and 19x cheaper than Anthropic's Claude Opus 4.7 to run the same benchmark — a striking data point in the global AI cost war.
Key Takeaways
DeepSeek permanently cut the price of its V4 Pro model by 75% , announced on Saturday, 24 May 2026 .
V4 Pro now ranks at the global frontier for cost-efficiency, according to benchmark firm Artificial Analysis .
Running the Artificial Analysis Intelligence Index costs US$268 with V4 Pro , versus roughly 12x more for OpenAI GPT-5.5 and 19x more for Anthropic Claude Opus 4.7 .
API pricing starts at US$0.0036 per 1 million cached input tokens and US$0.87 per 1 million output tokens.
The V4 generation includes two variants: the flagship V4 Pro and the lighter V4 Flash .
DeepSeek, the Hangzhou-based Chinese AI start-up, has made a 75 per cent promotional price cut on its flagship V4 Pro model permanent, propelling it to the top of global cost-efficiency rankings and leaving rivals OpenAI and Anthropic far behind on a value-per-dollar basis. The announcement was made on Saturday, 24 May 2026, roughly one month after the V4 model generation — comprising the V4 Pro and the lighter V4 Flash — was first released.
Top of the bang-for-buck leaderboard
According to third-party benchmark firm Artificial Analysis, the price reduction places V4 Pro at the global frontier for cost-efficiency, measured by the amount of 'intelligence' extractable per dollar spent. The methodology has gained traction in recent months as a global compute supply crunch has pushed prices higher for the most advanced models, making raw performance rankings less meaningful for budget-conscious developers and enterprises.
DeepSeek's official API pricing for V4 Pro now starts at as low as US$0.0036 per 1 million cached input tokens and US$0.87 per 1 million output tokens.
How it compares to OpenAI and Anthropic
Running Artificial Analysis's Intelligence Index benchmark — a composite test combining multiple performance metrics — costs just US$268 using V4 Pro. By contrast, OpenAI's GPT-5.5 costs approximately 12 times more to complete the same benchmark, while Anthropic's Claude Opus 4.7 costs roughly 19 times more, according to the benchmark firm's assessment.
The gap underscores a deliberate strategic divergence: where US frontier labs compete primarily on raw capability at premium price points, Chinese AI developers are increasingly competing on accessibility and cost.
Why it matters
The permanent price reduction signals that DeepSeek is not treating aggressive pricing as a short-term tactic but as a core market positioning strategy. For developers, startups, and enterprises running large inference workloads, the economics of choosing a model provider are shifting materially.
Amid ongoing constraints in advanced compute availability — particularly for companies outside the US supply chain — cost-efficient inference is emerging as a decisive competitive dimension in the global AI race.
What's next
The move is likely to intensify pricing pressure across the industry, with rivals potentially forced to respond with their own cuts or justify premium pricing through differentiated capabilities. Watch whether OpenAI and Anthropic adjust their API pricing tiers in the coming weeks, and whether DeepSeek's V4 Flash variant sees similar permanent reductions.
Point of View
Chinese developers effectively redefine the competition on terrain where they hold a structural advantage. The compute supply crunch tightening around non-US players makes this even sharper: if you can't access the latest Nvidia silicon at scale, you engineer your way to the frontier on efficiency. The real pressure now falls on mid-tier Western API providers caught between frontier capability they can't match and pricing they can't sustain.
NationPress
9 Jul 2026
Frequently Asked Questions
What is DeepSeek V4 Pro and why is it in the news?
DeepSeek V4 Pro is the flagship model in the Chinese AI start-up's latest V4 generation, which also includes the lighter V4 Flash variant. It made headlines after DeepSeek permanently cut its API price by 75%, placing it at the top of global cost-efficiency rankings according to benchmark firm Artificial Analysis.
How much cheaper is DeepSeek V4 Pro compared to OpenAI and Anthropic?
Running the Artificial Analysis Intelligence Index benchmark costs US$268 with DeepSeek V4 Pro. The same benchmark costs approximately 12 times more using OpenAI's GPT-5.5 and about 19 times more using Anthropic's Claude Opus 4.7, according to Artificial Analysis data.
What is the current API price for DeepSeek V4 Pro?
DeepSeek V4 Pro's official API pricing starts at US$0.0036 per 1 million cached input tokens and US$0.87 per 1 million output tokens following the permanent price reduction announced on 24 May 2026.
Why are cost-efficiency rankings becoming more important in AI?
A global compute supply crunch has driven up prices for the most advanced AI models, making raw performance comparisons less useful for many developers and enterprises. The 'intelligence-per-dollar' metric, as measured by firms like Artificial Analysis, has gained popularity as a more practical way to evaluate AI model value.
How does DeepSeek's pricing strategy differ from US AI companies?
DeepSeek and other Chinese AI developers are competing primarily on cost-efficiency and accessibility, while US frontier labs such as OpenAI and Anthropic focus on cutting-edge capabilities at premium price points. DeepSeek's decision to make its promotional discount permanent suggests aggressive pricing is now a long-term strategic pillar rather than a temporary tactic.
