DeepSeek Platform重要新发布2 个来源印证2026-09-10 → 2026-09-20
DeepSeek 发布 V4.1 Flash 并大幅降价与合并视觉能力
DeepSeek Launches V4.1 Flash with Price Cuts and Native Vision
多模态效果提升
EVENT SUMMARY
事件摘要
DeepSeek 推出 V4.1 Flash 模型，替代原有 V4 Flash 系列。新模型在性能、成本和速度上据称全面超越 V4 Pro，并原生集成视觉能力，不再需要独立的 vision-exp 模型。定价方面，V4.1 Flash 价格大幅下调，输入缓存命中低谷价降至 $0.003/M tokens（降幅约 57%），缓存未命中和输出 token 价格也有显著下降。此举使 Flash 成为集文本与视觉的单一低价模型，提升了多模态 API 的性价比标准。
ANALYST VIEW
中立分析
为什么重要
V4.1 Flash 缓存命中价成为市场最低价层级之一，可能引发新一轮 API 推理价格竞争；Flash 级模型原生支持多模态视觉正成为行业基线，加速旗舰能力下沉。
影响对象
开发者无需再为视觉任务单独切换模型，使用 Flash 模型的开发者将获得 9%-57% 不等的成本下降，缓存命中场景受益最大。
证据说明
来源为官方定价页 diff，直接证明了 V4.1 Flash 发布、新旧价格数值对比、表格结构变化和 Vision 特性标记。
不确定性
V4.1 Flash 的具体基准测试分数和与 V4 Pro 的量化对比数据未披露；视觉能力的详细技术规格及与原有 vision-exp 的性能差异需查阅其他文档。
SOURCES
2 个来源 · 14 条记录
DeepSeek Platform·pricing指向来源页采集于 2026-09-20
DeepSeek Flash 模型确认底层为 V4.1-Flash，旧模型名已退役
来源原文：MODEL VERSION: DeepSeek-V4.1-Flash. Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired, their requests are served by the DeepSeek-V4.1-Flash model and billed at the Flash price.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek V4.1 Flash 定价公布：输入缓存命中 $0.003/M tokens，大幅低于前代
来源原文：1M INPUT TOKENS (CACHE HIT) OFF-PEAK $0.003; 1M INPUT TOKENS (CACHE MISS) OFF-PEAK $0.15; 1M OUTPUT TOKENS OFF-PEAK $0.6
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek 推出 V4.1 Flash 模型，替代原 V4 Flash 系列
来源原文：Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired, their requests are served by the DeepSeek-V4.1-Flash model and billed at the Flash price.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek Flash 层级定价大幅下调
来源原文：1M INPUT TOKENS (CACHE HIT) OFF-PEAK: $0.003 (was $0.007); 1M INPUT TOKENS (CACHE MISS) OFF-PEAK: $0.15 (was $0.22); 1M OUTPUT TOKENS OFF-PEAK: $0.6 (was $0.66).
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek 推出 deepseek-flash 模型名称及 V4.1 Flash，退役旧 Flash 与 Flash Vision 模型
来源原文：Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired, their requests are served by the DeepSeek-V4.1-Flash model and billed at the Flash price.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek 产品线精简：从三个模型合并为两个
来源原文：MODEL: deepseek-flash(1), deepseek-v4-pro(2)... Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek V4.1 Flash 大幅降价，输入价格下降约 32%
来源原文：1M INPUT TOKENS (CACHE MISS): OFF-PEAK $0.15 (was $0.22), PEAK $0.3 (was $0.44)... 1M INPUT TOKENS (CACHE HIT): OFF-PEAK $0.003 (was $0.007), PEAK $0.006 (was $0.014)... 1M OUTPUT TOKENS: OFF-PEAK $0.6 (was $0.66), PEAK $1.2 (was $1.32)
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-14
DeepSeek 推出新版 V4.1 Flash 模型，取代 V4 Flash 与 Flash Vision Exp
来源原文：Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired, their requests are served by the DeepSeek-V4.1-Flash model and billed at the Flash price.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-11
DeepSeek 定价页面移除独立 Vision 模型行，视觉能力并入 Flash
来源原文：Features row now shows 'Vision ✓' for deepseek-flash, and the model list changed from three columns to two, removing the standalone deepseek-v4-flash-vision-exp column.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-11
DeepSeek V4.1 Flash 价格大幅下调，输入缓存命中降至 $0.003/M tokens
来源原文：1M INPUT TOKENS (CACHE HIT) OFF-PEAK $0.003 / PEAK $0.006; 1M INPUT TOKENS (CACHE MISS) OFF-PEAK $0.15 / PEAK $0.3; 1M OUTPUT TOKENS OFF-PEAK $0.6 / PEAK $1.2
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·pricing指向来源页采集于 2026-09-11
DeepSeek 发布 V4.1 Flash 模型，替代 V4 Flash 系列
来源原文：Use deepseek-flash as the model name. The legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted, but the corresponding models have been retired, their requests are served by the DeepSeek-V4.1-Flash model and billed at the Flash price.
https://api-docs.deepseek.com/quick_start/pricing
DeepSeek Platform·changelog指向来源页2026-09-10
DeepSeek API 伴随V4.1-Flash降价
来源原文：With the release of DeepSeek-V4.1-Flash, API prices have been reduced accordingly. For details, please refer to Models & Pricing.
https://api-docs.deepseek.com/updates
DeepSeek Platform·changelog指向来源页2026-09-10
DeepSeek API 价格下调
来源原文：With the release of DeepSeek-V4.1-Flash, API prices have been reduced accordingly. For details, please refer to Models & Pricing.
https://api-docs.deepseek.com/updates
DeepSeek Platform·changelog指向来源页2026-09-10
DeepSeek V4.1 Flash API 上线及旧模型路由变更
来源原文：DeepSeek V4.1 Flash is now available on the DeepSeek API with native multimodal support. Change the model name to deepseek-flash to call the latest V4.1 Flash model. The previous-generation models V4 Flash and V4 Flash Vision Exp have been retired... we plan to retire V4 Pro in an orderly manner. After 12:00 Beijing Time on September 14, 2026, all requests to deepseek-v4-pro will be routed to V4.1 Flash and billed at the V4.1 Flash price.
https://api-docs.deepseek.com/updates