DeepSeek: DeepSeek v4.1 Flash 1017

本文为原文前 6,000 字符的节选翻译，完整内容请查看原文。

deepseek/deepseek-v4.1-flash-1017

DeepSeek V4.1 Flash 1017 是 DeepSeek V4.1 Flash 的 2026 年 10 月 17 日快照版本，由 DeepSeek 以 deepseek-flash API 模型名称提供服务。V4.1 Flash 是 DeepSeek 开发的一种稀疏混合专家模型，也是首个基于该公司因果编码器-解码器（CED）架构构建的模型。它在输入时激活 8B 参数，输出时激活 16B 参数，骨干网络总参数量为 552B；这种非对称拆分使得每个 token 的计算量相对于模型总规模而言保持在较低水平。图像理解是该架构的原生功能，视觉和文本嵌入从预训练之初就进行了联合训练，而不是像早期的实验性 V4 Flash Vision Exp 那样在后期添加。

DeepSeek V4.1 Flash 1017 is the October 17, 2026 snapshot of DeepSeek V4.1 Flash, served by DeepSeek under the deepseek-flash API model name. V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company’s Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on output from a 552B-parameter backbone, an asymmetric split that keeps per-token compute low relative to the model’s total size. Image understanding is native to the architecture, with visual and text embeddings trained jointly from the start of pre-training rather than added afterward as in the earlier experimental V4 Flash Vision ExpOpens in new tab.

它适用于编码、终端和计算机使用代理，以及需要跨多个步骤运行至完成的长周期任务和长上下文分析。压缩的 KV 缓存将缓存内存减少到上一代 Flash 的四分之一左右，显著降低了代理工作负载的成本。DeepSeek 将其定位为 V4.1 系列中具有成本效益的层级，并报告称其在性能、速度和任务完成时间方面均超过了 V4 Pro。

It is suited for coding, terminal, and computer-use agents, along with long-horizon tasks that must run to completion across many steps and long-context analysis. Compressed KV caching cuts cache memory to roughly a quarter of the previous Flash generation, significantly reducing costs on agentic workloads. DeepSeek positions it as the cost-efficient tier of the V4.1 family and reports that it exceeds V4 ProOpens in new tab on performance, speed, and task completion time.

不同的公司托管同一个模型。OpenRouter 会根据您选择的路由模式（平衡（价格+速度）、Nitro（最快）或 Floor（最便宜））将您的请求路由到其中一家公司。

Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick — Balanced (price + speed), Nitro (fastest), or Floor (cheapest).

客户为该模型实际支付的平均价格，以及提供商发布的价格。缓存和折扣意味着实际支付的价格通常远低于标价。

The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.

吞吐量是模型写入的速度（每秒 token 数——越高越好）。延迟是总往返时间（越低越好）。TTFT 是首个 token 生成时间——即在看到任何内容出现之前需要等待的时间（越低越好）。

Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).

正常运行时间是过去 3 天内至少有一家提供商响应请求的百分比。可用性是推理成功服务的百分比时间。OpenRouter 会持续监控，并在某家提供商返回错误时使用次优提供商。

Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.

标准化评估的分数以及该模型在共享基准提示词上的输出。百分比越高越好——每个排名显示了该模型在评估来源中的位置。

Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.

向该模型发送最多流量的公共应用程序。这是了解真实生产工作负载的良好信号，也暗示了该模型最适合哪些用例。

Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.

该模型的 token 量和请求流量随时间的变化。

Token volume and request traffic to this model over time.

调用该模型的嵌入式代码。OpenRouter 的 API 与 OpenAI 兼容——大多数 SDK 只需更换基础 URL 即可工作。模型之间唯一改变的是下方的模型标识符（slug）。

Drop-in code to call this model. OpenRouter’s API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.

DeepSeek：DeepSeek Flash 版本

DeepSeek: DeepSeek Flash versions

目前无法通过欧盟区域内路由使用。区域内路由文档

Not currently available through EU in-region routing. In-region routing docs

使用 DeepSeek 构建的指南、食谱、博客文章和工具：OpenRouter 上的 DeepSeek V4.1 Flash 1017。

Guides, cookbooks, blog posts, and tools for building with DeepSeek: DeepSeek V4.1 Flash 1017 on OpenRouter.

P50，各提供商中的最佳值

P50, best across providers

当上游提供商发生错误时，如果您的请求过滤器允许，我们可以通过路由到另一个健康的提供商来恢复。您可以通过 Endpoints API 以编程方式访问各提供商的正常运行时间数据。了解更多关于我们的负载均衡和提供商路由选项的信息。

When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.

OpenRouter 上的 DeepSeek V4.1 Flash 1017 文档包括“DeepSeek V4 有视觉功能吗？”博客文章。快速入门涵盖了 API 密钥和您的第一个请求，API 参考列出了每个参数。

The DeepSeek V4.1 Flash 1017 documentation on OpenRouter includes the Does DeepSeek V4 Have Vision? blog post. The quickstart covers API keys and your first request, and the API reference lists every parameter.

DeepSeek V4.1 Flash 1017 的价格为每百万输入 token 0.08 美元，每百万输出 token 0.50 美元，缓存读取费率为每百万 token 0.008 美元。

DeepSeek V4.1 Flash 1017 costs $0.08/M input tokens and $0.50/M output tokens, with separate rates for Cache Read at $0.008/M tokens.

DeepSeek V4.1 Flash 1017 拥有 1,048,576 个 token 的上下文窗口。它支持最多 131,072 个补全 token。

DeepSeek V4.1 Flash 1017 has a 1,048,576 token context window. It supports up to 131,072 completion tokens.

是的。DeepSeek V4.1 Flash 1017 接受用于函数调用的 tools 和 tool_choice。它还通过 response_format 中的 JSON 模式支持结构化输出。

Yes. DeepSeek V4.1 Flash 1017 accepts tools and tool_choice for function calling. It also supports structured outputs via a JSON schema in response_format.

DeepSeek V4.1 Flash 1017 接受文本和图像作为输入，并返回文本。

DeepSeek V4.1 Flash 1017 accepts text and images as input and returns text.

DeepSeek V4.1 Flash 1017 由 OpenRouter 上的 8 家提供商提供服务：DeepInfra、inference.net、Morph、Cloudflare、DeepSeek、Relace、Baseten 和 Parasail。请求会被路由到最佳可用提供商，并自动故障转移到其他提供商，您可以使用提供商路由来固定或排除提供商。

DeepSeek V4.1 Flash 1017 is served by 8 providers on OpenRouter: DeepInfra, inference.net, Morph, Cloudflare, DeepSeek, Relace, Baseten, and Parasail. Requests are routed to the best available provider, with automatic failover to the others, and you can pin or exclude providers with provider routing.

DeepSeek V4.1 Flash 1017 发布于 2026 年 10 月 17 日。

DeepSeek V4.1 Flash 1017 was released on October 17, 2026.

此模型始终重定向到 DeepSeek Pro 系列中的最新模型。

This model always redirects to the latest model in the DeepSeek Pro family.

此模型始终重定向到 DeepSeek Flash 系列中的最新模型。

This model always redirects to the latest model in the DeepSeek Flash family.

DeepSeek V4.1 Flash 是 DeepSeek 开发的一种稀疏混合专家模型，也是首个基于该公司因果编码器-解码器（CED）架构构建的模型。它在输入时激活 8B 参数，输出时激活 16B 参数，骨干网络总参数量为 552B；这种非对称拆分使得每个 token 的计算量相对于模型总规模而言保持在较低水平。图像理解是该架构的原生功能，视觉和文本嵌入从预训练之初就进行了联合训练，而不是像早期的实验性 V4 Flash Vision Exp 那样在后期添加。

DeepSeek V4.1 Flash is a sparse mixture-of-experts model from DeepSeek, and the first built on the company’s Causal Encoder-Decoder (CED) architecture. It activates 8B parameters on input and 16B on output from a 552B-parameter backbone, an asymmetric split that keeps per-token compute low relative to the model’s total size. Image understanding is native to the architecture, with visual and text embeddings trained jointly from the start of pre-training rather than added afterward as in the earlier experimental V4 Flash Vision Exp.

DeepSeek V4.1 Flash 是 DeepSeek 开发的一种稀疏混合专家模型，也是首个基于该公司因果编码器-解码器（CED）架构构建的模型。它在输入时激活 8B 参数，输出时激活 16B 参数，骨干网络总参数量为 552B；这种非对称拆分使得每个 token 的计算量相对于模型总规模而言保持在较低水平。