DeepSeek V4.1 Flash begins internal testing, and you can use it directly by simply changing the model ID.

动察 Beating AI Flash News: DeepSeek has begun internal testing of the V4.1 Flash intermediate version. An official WeChat group notification stated that the new model adopts a new architecture, natively supports multimodal capabilities, and offers stronger performance, faster speed, and lower costs. Current pricing remains the same as V4 Flash, with a maximum of 20 concurrent requests per account.

Users can now directly call it. The base_url remains unchanged; simply replace the original model ID with deepseek-v4.1-flash-expires-on-0910.