DeepSeek V4 Flash 0731 智能性能与价格分析（Artificial Analysis）
Source: <https://artificialanalysis.ai/models/deepseek-v4-flash> | 2026-07-31
Author: Artificial Analysis | Category: models | Quality Score: 4/5
Tags: deepseek, benchmark, pricing, open-weights, llm
English Summary
Artificial Analysis rates DeepSeek V4 Flash 0731 (Reasoning, Max Effort) at Intelligence Index 50 (rank 3/101 globally), well above the median of 25. Pricing is competitive: $0.14/1M input tokens, $0.28/1M output tokens (median: $0.43/$1.20). Cache hit price is $0.003/1M (rank 1/101, 98% below median). 1M token context window, text-in/text-out. Generated 210M tokens during Intelligence Index evaluation (very verbose vs median 100M).
中文概要
Artificial Analysis 对 DeepSeek V4 Flash 0731（Reasoning, Max Effort）的综合评测：智能指数 50（全网第 3/101），远超同类型模型中位数 25价格极具竞争力：输入 $0.14/1M tokens输出 $0.28/1M tokens，均低于市场中位数（0.43/1.20）缓存命中价格 $0.003/1M（全网最低，下降 98%）上下文窗口 1M tokens，支持文本输入输出在智能指数评测中产生 210M tokens，冗余度高于中位数的 100M