DeepSeek launches smaller but faster AI model

DeepSeek released a new smaller model on Thursday that it says can outperform some flagship systems while cutting inference costs, reports Caixin. This comes as Chinese AI developers increasingly compete on both model capability and price.

The new DeepSeek-V4.1-Flash is a 552-billion-parameter mixture-of-experts model with native multimodal visual understanding. DeepSeek said the model uses a new architecture designed to deliver higher performance, faster inference, higher throughput and easier scaling to larger models.

DeepSeek said V4.1-Flash outperformed domestic rivals including Moonshot AI’s Kimi K3 and Z.AI’s GLM-5.3 on benchmarks such as Terminal-Bench 3.0, DeepSWE v1.1, CyberGym and Automation-Bench. It also beat Anthropic’s Opus 5 and OpenAI’s GPT-5.6 Sol on some tests, according to the company.