DeepSeek V4 Flash 0731 (Reasoning, Max Effort) is a large-scale Mixture of Experts (MoE) model released on July 31, 2026, featuring 284 billion total parameters with only 13 billion active parameters during inference. Designed for high-performance reasoning tasks, the model supports text-based input and output with a significant 1 million token context window. On the Artificial Analysis Intelligence Index, the model achieved a score of 34, significantly outperforming the median score of 18 for comparable models. Technically, the model is characterized by its high verbosity, generating 240 million tokens during intelligence evaluations compared to a median of 140 million. Performance metrics highlight exceptional speed, with an output rate of approximately 222 tokens per second and a competitive time to first token (TTFT) of 1.02 seconds. While the model is noted for being somewhat more expensive than similar open-weight models, its API pricing is set at $0.44 per 1 million input tokens and $1.32 per 1 million output tokens. This results in an average cost of $0.22 per task when evaluated on the Intelligence Index. Released under the MIT license, the model is optimized for complex problem-solving through extended chain-of-thought reasoning.