New Deepseek Flash model matches OpenAI's GPT-5.6 Luna at roughly 60 percent lower cost
Deepseek's updated Flash V4 model scores 50 on the Artificial Analysis Intelligence Index, nearing OpenAI's GPT-5.6 Luna performance while costing roughly 60 percent less per task.
Deepseek has released an update to its Flash V4 model, designated 0731, which significantly improves its benchmark performance. The model now scores 50 on the Artificial Analysis Intelligence Index, placing it within one point of OpenAI's GPT-5.6 Luna.
The primary advantage lies in pricing efficiency. According to the report, the Deepseek model delivers comparable performance at approximately 60 percent lower cost per task compared to the OpenAI alternative. This positions it as a competitive option for developers seeking high performance without premium pricing.
This development highlights the ongoing competition in the large language model sector, where efficiency and cost-effectiveness are becoming key differentiators alongside raw capability. Deepseek's progress suggests that lower-cost architectures can achieve parity with leading proprietary models on specific benchmarks.
The Artificial Analysis Intelligence Index serves as a metric for evaluating model capabilities. Deepseek's jump of ten points indicates a substantial optimization in the latest update, reinforcing the company's reputation for delivering high-value AI tools.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.