The DeepSeek-V4.1-Flash-20261017 model, released on October 17, 2026, has secured the top position among open models on the Arena.ai Code Arena WebDev leaderboard. The model achieved a score of 1670 points, placing it 12th overall in the rankings for agentic and front-end coding tasks.
This update outperformed the previous leader, Kimi K3 (Max), while offering a cost advantage of approximately 10 times lower. The model utilizes a Mixture-of-Experts architecture with 552 billion backbone parameters and 196 billion Engram parameters, activating 8 billion during prefill and 16 billion during decoding.
Additional technical features include a Causal Encoder-Decoder structure, native vision support, and a 1 million token context window. The model weights are available on Hugging Face under an MIT license.
For developers opting for API access, the service is priced at $0.30 per million input tokens and $1.20 per million output tokens. Hangzhou DeepSeek continues to focus on developing cost-efficient models for agentic coding workflows.
