Chinese AI startup DeepSeek has launched V4-Flash, which research firm Artificial Analysis says is the world's cheapest well-known AI model to run, undercutting rivals including Anthropic's Claude Fable 5 and OpenAI's GPT-5.6.
DeepSeek, which Reuters reported is preparing for a potential initial public offering (IPO), unveiled V4-Flash on Friday as it seeks to regain momentum through ultra-low-cost AI models.
The startup gained global attention earlier in 2025 when its R1 model triggered a selloff in technology stocks and sparked debate over the heavy investments US companies were making in artificial intelligence, according to the report.
ALSO READ | One Founder, AI Workforce: The New Formula For Million-Dollar Startups
According to Artificial Analysis, V4-Flash costs $0.14 per million input tokens and $0.28 per million output tokens. A token is a unit of data used to measure AI model usage.
The San Francisco-based research firm estimated V4-Flash's average cost at 3 cents per benchmark test, compared with 86 cents for Chinese rival Moonshot AI's Kimi K3, $1.86 for OpenAI's GPT-5.6 Sol and $3.15 for Anthropic's Claude Fable 5.
The benchmark provides a more meaningful comparison than pricing alone because it accounts for the amount of data a model processes and generates to complete a task. Even models with lower headline prices can prove more expensive if they require significantly more computing steps to produce an answer.
Artificial Analysis gave V4-Flash a score of 50 out of 100 on its Intelligence Index, which measures AI models across nine benchmarks covering reasoning, coding and workplace-style tasks.
The model matched Google's Gemini 3.6 Flash and trailed Meta's Muse Spark 1.1 and Z.AI's GLM-5.2 by just one point, highlighting its competitive performance despite its significantly lower operating cost.
ALSO READ | Indian IT's AI Story Is Changing, But BofA Flags A New Earnings Risk
DeepSeek, once the dominant name in China's AI boom, now faces intensifying competition from rivals including ByteDance, Alibaba, Moonshot AI, MiniMax and Z.AI, as Chinese developers race alongside US technology companies to offer more capable and affordable AI models.
Separately on Monday, Alibaba unveiled Qwen3.8-Max, its largest and most powerful AI model to date, matching the scale of a model introduced by Moonshot AI last month.
DeepSeek is also developing V4-Pro, a more powerful version of its latest model, although the company has not announced a launch date.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.