DeepSeek V4.1 Flash 1017
工具调用推理缓存结构化输出
官方价格
$0.15 / $0.6
DeepSeek · off-peak · 输入 / 输出 · 每百万 token · Official off-peak prices; peak input $0.30 and output $1.20 per 1M tokens. Cache hits $0.003 off-peak / $0.006 peak.
最低输入价
$0.036
Wafer
上下文
1M
比较服务商
每个点是一家服务商。从左到右是每百万 token 的输入价,从下到上是输出价。 离其余报价太远的价格没有画在这张图上。
基础价格
43最新价格检查于 Nov 18, 2026。
服务商价格
6金额均以美元计价。悬停或点击带下划线的金额,可查看价格换算明细。
DR 是服务商网站的 Ahrefs 域名评级,范围 0 到 100。域名年龄是该网站的注册时间。这些只说明公开痕迹,不能说明服务商是否会持续运营。域名评级低于 10 的报价不列出。
未验证报价(3)
金额均以美元计价。悬停或点击带下划线的金额,可查看价格换算明细。
DR 是服务商网站的 Ahrefs 域名评级,范围 0 到 100。域名年龄是该网站的注册时间。这些只说明公开痕迹,不能说明服务商是否会持续运营。域名评级低于 10 的报价不列出。
DeepSeek V4.1 Flash 1017 · 已公布价格
| Period | Input / 1M | Cache read / 1M | Output / 1M |
|---|---|---|---|
| Off-peak | $0.15 | $0.003 | $0.60 |
| Peak | $0.30 | $0.006 | $1.20 |
DeepSeek V4.1 Flash 1017 · 模型规格
| Field | Official API |
|---|---|
| Snapshot | DeepSeek-V4.1-Flash-1017 · October 17, 2026 |
| API alias | deepseek-flash |
| Context / maximum output | 1M / 384K tokens |
| Input | Text and images |
| Features | JSON output, tool calls, Responses API, Anthropic API |
| Backbone parameters | 552B; additional Engram memory is separate |
| Active parameters per token | 8B prefill / 16B decode |
| Engram conditional memory | 196B, sparsely accessed through token-based lookup |
| Official checkpoint | deepseek-ai/DeepSeek-V4.1-Flash-1017 |
DeepSeek V4.1 Flash 1017 · 权重许可证
| 模型权重 | 协议 |
|---|---|
| 开放权重 | MIT |
缓存与处理模式
关于 DeepSeek V4.1 Flash 1017
It is the October 17, 2026 snapshot replacing the September 10 version. DeepSeek kept the deepseek-flash API alias and token prices unchanged.
Per million tokens, official off-peak prices are $0.15 uncached input, $0.003 cache hits and $0.60 output. Peak prices are $0.30, $0.006 and $1.20. Peak hours are Monday–Friday 01:00–04:00 and 06:00–10:00 UTC, excluding Chinese public holidays.
No. DeepSeek lists 1M context and up to 384K output on its API. Hosted endpoints can publish different output limits and tariffs. Check the selected provider instead of applying the publisher limit to every channel.
DeepSeek publishes the 1017 model weights on Hugging Face under the MIT license. Self-hosting hardware and running costs are separate from hosted API token prices.
One million uncached input tokens plus 100,000 output tokens cost $0.21 off-peak or $0.42 at peak using the official tariff. These are arithmetic examples; token use, cache eligibility and provider-specific rates change the actual bill.
来源与注脚
- [1]
- [2]
- [3]
- [4]
- [5]
输入、输出和缓存读取的单位是每百万 token 的美元价格。速度是每秒输出 token 数。可用率是来源报告的在线比例。




