DeepSeek-V4-Flash single query costs around $0.03, surging to the forefront of cost-effectiveness in the model.

According to Perceive Beating monitoring, in the latest Artificial Analysis evaluation, the intelligent index of DeepSeek-V4-Flash official version has reached around 50 points. Calculated based on the DeepSeek official API and an approximately 99% cache hit rate, the average cost per task is around $0.03.

As a result, it has entered the "Pareto Frontier" of intelligence and cost: among all tested models, hardly any model can be found that is both stronger and cheaper than it.

For a higher capability, costs usually increase significantly. To lower the price, model performance would decrease. DeepSeek-V4-Flash is currently teetering on the most cost-effective balance between the two.