DeepSeek's New AI Model is World's Cheapest to Run, Research Firm Claims

Chinese artificial intelligence startup DeepSeek has released its V4-Flash model, which research firm Artificial Analysis reports is the least expensive among well-known global AI models to operate. This move aims to regain momentum for the company, which is reportedly preparing for a potential IPO.

“

A version of Chinese startup DeepSeek's flagship AI model is by far the least expensive to run on benchmark tests among well-known models globally and more than 100 times cheaper to run than Anthropic's Claude Fable 5, according to a research firm.

DeepSeek's V4-Flash model charges $0.14 per million input tokens and $0.28 per million output tokens. In benchmark tests, this makes it more than 100 times cheaper to run than Anthropic's Claude Fable 5. Artificial Analysis estimated the average cost for V4-Flash at just 3 cents per test, a stark contrast to $1.86 for OpenAI's GPT-5.6 Sol and $3.15 for Claude Fable 5. This comparison offers a more accurate measure of value by considering the data processing required for a task, not just the headline price.

DeepSeek's R1 model previously gained attention in early 2025 for its ultra-low-cost AI alternatives, sparking discussions about the substantial spending by U.S. companies on AI development. The company faces intense competition from domestic rivals like Moonshot AI, MiniMax, and Z.AI, as well as tech giants such as ByteDance and Alibaba, all vying for global market share.

“

DeepSeek's V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens, according to research firm Artificial Analysis.

Despite its cost advantage, DeepSeek's V4-Flash model scored 50 out of 100 on Artificial Analysis's Intelligence Index, matching Google's Gemini 3.6 Flash. This score is lower than Meta's Muse Spark 1.1 and Z.AI's GLM-5.2. Competitors like Moonshot's Kimi K3 scored 57, while Anthropic's Claude Opus 5, Fable 5, and OpenAI's GPT-5.6 achieved scores nine or more points higher. DeepSeek is also developing a more powerful version, V4-Pro, though a release date has not been announced. Meanwhile, Alibaba recently unveiled its Qwen3.8-Max, its most capable AI model to date.

“

The comparison provides a more realistic measure of value than pricing alone because it accounts for the amount of data a model must process and generate to complete a task.