Chinese AI startup DeepSeek has launched DeepSeek V4-Flash, its latest flagship AI model, which has emerged as the lowest cost model to run among leading global AI systems, according to research firm Artificial Analysis.

The company officially released DeepSeek V4-Flash on Friday as part of its strategy to strengthen its position in the competitive AI market by offering high performance at significantly lower costs.

Earlier this year, DeepSeek gained worldwide attention after the release of its R1 model. The model triggered a sharp selloff in global technology stocks and raised questions about the high spending by U.S. technology companies on artificial intelligence.

According to Artificial Analysis, DeepSeek V4-Flash charges $0.14 per million input tokens and $0.28 per million output tokens. Tokens are units of data used to measure AI usage.

According to a Reuters report, the launch marks DeepSeek’s latest effort to regain momentum by focusing on ultra low cost AI models. Reuters also reported that DeepSeek is preparing for a potential initial public offering (IPO), according to sources familiar with the matter.

The research firm estimated that the model costs an average of just $0.03 per benchmark test. By comparison, China’s Moonshot AI Kimi K3 costs $0.86 per test, OpenAI’s GPT-5.6 Sol costs $1.86, while Anthropic’s Claude Fable 5 costs $3.15. This makes DeepSeek’s latest model more than 100 times cheaper to run than Claude Fable 5.

Artificial Analysis noted that benchmark testing provides a more realistic measure of value than token pricing alone. The evaluation considers the amount of data a model processes and generates to complete a task. As a result, models with lower listed prices can still become more expensive if they require additional processing steps.

DeepSeek, once the leading name in China’s AI startup ecosystem, now faces increasing competition from domestic rivals. Companies including Moonshot AI, MiniMax, Z.AI, ByteDance, and Alibaba are rapidly expanding their AI offerings. At the same time, these firms are competing with U.S. technology companies for global enterprise adoption by providing more affordable AI solutions.

On the Artificial Analysis Intelligence Index, DeepSeek V4-Flash scored 50 out of 100. The index combines results from nine benchmarks covering coding, reasoning, and workplace related tasks.

The score matches Google’s Gemini 3.6 Flash and is one point behind Meta’s Muse Spark 1.1 and Z.AI’s GLM-5.2. However, Moonshot AI’s Kimi K3 achieved 57 points, while Anthropic’s Claude Opus 5, Claude Fable 5, and OpenAI GPT-5.6 scored at least nine points higher than DeepSeek’s latest model.

Meanwhile, DeepSeek is also preparing a more advanced version of the model called V4-Pro. The company has not announced an official release date.

Separately, Alibaba unveiled Qwen3.8-Max on Monday, describing it as its largest and most capable AI model to date. The model closely matches the scale of a competing offering launched last month by Moonshot AI.

As competition among Chinese AI developers continues to intensify, DeepSeek V4-Flash is positioning itself as a cost effective option for enterprises seeking affordable AI deployment without significantly compromising performance.

Related post

View all