DeepSeek senior researcher Victor Chen announced on X that the company's newly released DeepSeek-V4-Pro model will be offered at a huge discount over the next week, a move that threatens to unleash an AI platform price war just as Anthropic, OpenAI, and Google are rolling out newer, more expensive models.
"Second price drop in two days! On top of the base 75% off, stack an extra 90% discount for cache hits. That brings it down to just 0.003625 USD/0.025 RMB per 1M input tokens with cache hit ~ 🎉💰 Go wild and have fun ~," Chen wrote in a post on X late Sunday night.
He added, "Just a heads-up: the cache discount is permanent, while the base 75% off promo runs until May 5, so make the most of it while you can!"
The long-awaited V4 model was released at the end of last week, ending months of silence from one of China's most closely watched AI labs and arriving a year after its R1 release sparked U.S. equity market turmoil.
The open-source model comes in the V4 Flash and V4 Pro series, with DeepSeek saying its V4 "leads all current open models, trailing only Gemini-3.1-Pro."
DeepSeek's hefty discount is aimed at luring developers, startups, and enterprise users away from expensive U.S. models like those from OpenAI, Anthropic, and Google by offering lower prices, easier access, open-source availability, and a 1-million-token context window.
X user thehype pointed out that the Chinese AI lab's discount "is starting a price war in the AI market," adding:
they just slashed input cache prices to 1/10th of what they already were.
and there's a separate 75% off promo on v4-pro running until may 5th.
but even ignoring the sales – the normal api prices tell the story. output per 1M tokens (real weighted avg, no discounts):
- gpt-5.5: $30.21
- claude opus 4.7: $25.00
- deepseek v4-pro: $1.73
that's ~17x cheaper than gpt-5.5 and ~14x cheaper than opus 4.7.
now add the 75% promo: deepseek output drops to $0.87/M. that's 35x cheaper than gpt-5.5 and 29x cheaper than opus 4.7.
and the benchmarks? v4-pro isn't that far behind. artificial analysis intelligence index:
- gpt-5.5: 60
- claude opus 4.7: 57
- deepseek v4-pro: 52
13% lower score. 35x lower price.
after releasing v4 on open weights (mit license, free to self-host), deepseek is now aggressively competing on cloud api pricing too. own both ends of the market.
it's a dangerous game. when a model is 87% as capable at 6% of the cost, "we're better" stops being a pitch
ai is starting to commodify. the price war has begun.
Another X user asked what DeepSeek's actual motive is behind the price drop:
AI price war it is.