feat(provider): add DeepSeek V4.1 Flash and retirement notices (#267)

Add `deepseek-flash` with native vision, thinking mode, and tool calling.
Keep existing model entries and settings, with retirement and migration guidance in the model picker.

Related to #265

Author: Vizards

README.md

@@ -21,34 +21,34 @@
   <img src="resources/screenshots/01-picker.png" alt="DeepSeek V4 Flash, Flash Vision Exp, and Pro in the Copilot Chat model picker, alongside the Thinking Effort menu" width="800">
 </p>
 
-Love DeepSeek's price-performance but don't want to give up GitHub Copilot's agent mode, tool calling, and polished UI? This extension adds **DeepSeek V4 Flash, Pro, and Flash Vision Exp** to the Copilot Chat model selector — with **native vision or Vision Proxy**, **thinking mode**, and your own API key.
+Love DeepSeek's price-performance but don't want to give up GitHub Copilot's agent mode, tool calling, and polished UI? This extension adds **DeepSeek V4.1 Flash and other DeepSeek models** to the Copilot Chat model selector — with **native vision or Vision Proxy**, **thinking mode**, and your own API key.
 
 ## Why this extension?
 
 - **Don't replace Copilot — power it up.** No new sidebar, no new chat UI to learn. Just a new model in the picker you already use.
 - **Agent mode, tool calling, instructions, MCP, skills — all of it still works.** Copilot's entire stack, now running on DeepSeek.
-- **Two ways to work with images.** Flash Vision Exp receives image attachments natively. Flash and Pro keep their existing text context while a configurable Vision Proxy turns images into descriptions.
+- **Two ways to work with images.** V4.1 Flash and Flash Vision Exp receive image attachments natively. V4 Flash and Pro keep their existing text context while a configurable Vision Proxy turns images into descriptions.
 - **BYOK, pay DeepSeek directly.** Your API key, your bill, your rate limits. Stored in the OS keychain, never on disk.
 
 ## Features
 
-### Three DeepSeek V4 models in the model picker
-Flash, Pro, and the experimental Flash Vision Exp appear alongside other models in Copilot Chat's model selector. All three support DeepSeek's long context, tool calling, and configurable thinking effort.
+### DeepSeek models in the model picker
+The picker includes V4.1 Flash and three legacy V4 entries, with long context, tool calling, and configurable thinking effort.
 
 ### Native Vision and Vision Proxy
 Choose the image path that fits the conversation:
 
-- **DeepSeek V4 Flash Vision Exp** handles image attachments as native multimodal input, without Vision Proxy. It is exposed as a separate experimental model and does not silently fall back when the configured API endpoint does not support its model ID.
-- **DeepSeek V4 Flash and Pro** use Vision Proxy: an image-capable model first describes each attachment, then the main DeepSeek model receives the description with the conversation. Auto mode selects Flash Vision Exp when available, while an explicitly configured VS Code model or API endpoint remains supported.
+- **DeepSeek V4.1 Flash and Flash Vision Exp** handle image attachments natively, without Vision Proxy.
+- The **DeepSeek V4 Flash and Pro** entries use Vision Proxy: an image-capable model first describes each attachment, then the main DeepSeek model receives the description with the conversation. Auto mode selects Flash Vision Exp when available, while an explicitly configured VS Code model or API endpoint remains supported.
 
-Avoid switching models mid-chat just to inspect an image if DeepSeek prefix-cache reuse matters. Start the conversation with Flash Vision Exp for native vision, or stay on Flash/Pro and let Vision Proxy preserve the main model choice.
+[Context caching](https://api-docs.deepseek.com/guides/kv_cache/) requires a full match with a cached prefix. Choose V4.1 Flash for new conversations and plan when to migrate existing ones.
 
 <p align="center">
   <img src="resources/screenshots/03-vision.png" alt="Dropping an image into Copilot Chat and DeepSeek responding to it via the vision proxy" width="800">
 </p>
 
-### Thinking Mode with Reasoning Effort Control
-Full support for DeepSeek V4's `reasoning_content`. Flash, Pro, and Flash Vision Exp offer `none` (off), `low` (light reasoning), `high` (balanced, default), and `max` (deep reasoning for hard agent tasks), matching the effort levels implemented by the official API.
+### Thinking Mode and Effort Control
+Supports DeepSeek's thinking mode and `reasoning_content`. `none` disables thinking mode; `low`, `high` (default), and `max` set the thinking effort.
 
 ### Inherits Every Copilot Capability
 Because this plugs into Copilot's native provider API, you get the full stack for free:
@@ -86,30 +86,31 @@ Install from the registry used by your editor:
 
 1. Run **DeepSeek: Set API Key** from the Command Palette (`Cmd+Shift+P`)
 2. Paste your key or compatible provider token (official DeepSeek keys usually start with `sk-`)
-3. Open Copilot Chat, click the model picker, and choose **DeepSeek V4 Flash**, **DeepSeek V4 Pro**, or **DeepSeek V4 Flash Vision Exp**
+3. Open Copilot Chat, click the model picker, and choose **DeepSeek V4.1 Flash**
 4. That's it — chat away
 
 ## Models
 
-| Model | Image Handling | Thinking Effort | Best For |
+| Model Entry | Image Handling | Thinking Effort | Official API Status |
 |---|---|---|---|
-| **DeepSeek V4 Flash** | Vision Proxy | `none` / `low` / `high` / `max` | Fast everyday coding, quick edits, cheap iteration |
-| **DeepSeek V4 Pro** | Vision Proxy | `none` / `low` / `high` / `max` | Complex refactors, agent tasks, deep reasoning |
-| **DeepSeek V4 Flash Vision Exp** | Native image input | `none` / `low` / `high` / `max` | Direct, experimental image understanding with fast reasoning |
+| **DeepSeek V4.1 Flash** | Native image input | `none` / `low` / `high` / `max` | Recommended for new conversations |
+| **DeepSeek V4 Flash** | Vision Proxy | `none` / `low` / `high` / `max` | Model retired; legacy name still accepted |
+| **DeepSeek V4 Pro** | Vision Proxy | `none` / `low` / `high` / `max` | Retires on September 14, 2026 at 12:00 Beijing time |
+| **DeepSeek V4 Flash Vision Exp** | Native image input | `none` / `low` / `high` / `max` | Model retired; legacy name still accepted |
 
-All three support optional thinking mode, tool calling, and 1M token context. Flash Vision Exp is experimental; custom API endpoints and compatible providers must expose its configured model ID to use it directly.
+All four entries support thinking mode, tool calling, and 1M token context. See the official [Models & Pricing](https://api-docs.deepseek.com/quick_start/pricing/) documentation for retirement, routing, and billing details.
 
 ## Settings
 
 | Setting | Default | Description |
 |---|---|---|
 | `deepseek-copilot.baseUrl` | `https://api.deepseek.com` | API endpoint — change for self-hosted / proxied deployments |
 | `deepseek-copilot.requestHeaders` | `{}` | Custom headers for chat completions. [Configuration guide](https://github.com/Vizards/deepseek-v4-for-copilot/blob/main/docs/settings/request-headers.en.md) |
-| `deepseek-copilot.maxTokens` | `0` | Max output tokens (`0` = no limit). Useful for cost control |
-| `deepseek-copilot.modelIdOverrides` | prefilled official ID map | API model IDs to send for DeepSeek V4 Flash, Pro, and Flash Vision Exp. Change only for compatible third-party APIs with different model names |
+| `deepseek-copilot.maxTokens` | `0` | Max output tokens (`0` = API default). Useful for cost control |
+| `deepseek-copilot.modelIdOverrides` | prefilled official ID map | API model IDs to send for V4.1 Flash and the three legacy V4 entries. Change only for compatible third-party APIs with different model names |
 | `deepseek-copilot.debugMode` | `minimal` | Diagnostic mode: `minimal` for token usage only, `metadata` for privacy-preserving logs, or `verbose` for full request dumps and pipeline snapshots under extension global storage. Full dumps may include sensitive prompt text, tool schemas, file snippets, and image descriptions. Use `DeepSeek: Open Request Dumps Folder` to open the dump location |
-| `deepseek-copilot.visionModel` | *(auto)* | Vision Proxy used by Flash and Pro. Auto mode selects Flash Vision Exp when available; configure another VS Code model or API endpoint with `DeepSeek: Configure Vision Proxy` |
-| `deepseek-copilot.visionPrompt` | *(built-in)* | Prompt used by Flash/Pro's Vision Proxy to describe image attachments. It does not affect native Flash Vision Exp requests |
+| `deepseek-copilot.visionModel` | *(auto)* | Vision Proxy used by V4 Flash and Pro. Auto mode selects Flash Vision Exp when available; configure another VS Code model or API endpoint with `DeepSeek: Configure Vision Proxy` |
+| `deepseek-copilot.visionPrompt` | *(built-in)* | Prompt used by V4 Flash/Pro's Vision Proxy to describe image attachments. It does not affect native V4.1 Flash or Flash Vision Exp requests |
 | `deepseek-copilot.experimental.stabilizeToolList` | `false` | Experimental. Tries to pre-activate VS Code/Copilot virtual tools so the DeepSeek API `tools` parameter is more complete and stable across turns. May improve context-cache hit rate when enabled tools change between turns. Can increase input tokens because more function definitions may be included; cache-hit input tokens are cheaper but still count toward usage. Usually leave it off with 64 or fewer enabled tools unless the tool list still changes across turns; do not enable it with more than 128 enabled tools |
 
 Thinking Effort is configured from Copilot Chat's model picker for each DeepSeek model.
@@ -119,6 +120,7 @@ Example `settings.json` override for compatible API proxies:
 ```json
 {
   "deepseek-copilot.modelIdOverrides": {
+    "deepseek-flash": "your-v4.1-flash-model-id",
     "deepseek-v4-flash": "your-flash-model-id",
     "deepseek-v4-pro": "your-pro-model-id",
     "deepseek-v4-flash-vision-exp": "your-vision-model-id"

README.zh-cn.md

@@ -21,34 +21,34 @@
   <img src="resources/screenshots/01-picker.png" alt="DeepSeek V4 Flash、Flash Vision Exp 和 Pro 出现在 Copilot Chat 模型选择器中，并展示思考深度菜单" width="800">
 </p>
 
-喜欢 DeepSeek 的性价比，但不想放弃 GitHub Copilot 的 Agent 模式、工具调用和成熟的交互体验？本扩展将 **DeepSeek V4 Flash、Pro 和 Flash Vision Exp** 直接接入 Copilot Chat 模型选择器，支持**原生视觉或视觉代理**、**思考模式**，并使用你自己的 API Key。
+喜欢 DeepSeek 的性价比，但不想放弃 GitHub Copilot 的 Agent 模式、工具调用和成熟的交互体验？本扩展将 **DeepSeek V4.1 Flash 等模型**直接接入 Copilot Chat 模型选择器，支持**原生视觉或视觉代理**、**思考模式**，并使用你自己的 API Key。
 
 ## 为什么选这个扩展？
 
 - **不是替换 Copilot，而是增强它。** 没有新的侧边栏，没有新的聊天界面需要学习。只是在你已经在用的模型选择器中多了一个选项。
 - **Agent 模式、工具调用、Instructions、MCP、Skills——全部正常运作。** Copilot 的完整能力栈，现在跑在 DeepSeek 上。
-- **两种图片处理方式。** Flash Vision Exp 会原生接收图片附件；Flash 和 Pro 则保留原有文本上下文，由可配置的视觉代理将图片转换为文字描述。
+- **两种图片处理方式。** V4.1 Flash 和 Flash Vision Exp 会原生接收图片附件；V4 Flash 和 Pro 则保留原有文本上下文，由可配置的视觉代理将图片转换为文字描述。
 - **需自行提供 API Key，直接向 DeepSeek 付费。** 你的 API Key，你的账单，你的速率限制。密钥存储在操作系统密钥链中，不会以明文形式写入磁盘。
 
 ## 功能特性
 
-### 三种 DeepSeek V4 模型出现在模型选择器中
-Flash、Pro 和实验性的 Flash Vision Exp 会与其他模型并列出现在 Copilot Chat 的模型选择器中。三者均支持 DeepSeek 的长上下文、工具调用和可配置的思考深度。
+### DeepSeek 模型出现在模型选择器中
+选择器包含 V4.1 Flash 和三个旧 V4 入口，支持长上下文、工具调用和可配置的思考强度。
 
 ### 原生视觉与视觉代理
 可以根据对话需要选择不同的图片处理路径：
 
-- **DeepSeek V4 Flash Vision Exp** 将图片附件作为原生多模态输入处理，不经过视觉代理。它是一个独立暴露的实验模型；如果当前 API 端点不支持其模型 ID，插件不会静默降级。
-- **DeepSeek V4 Flash 和 Pro** 使用视觉代理：先由支持图片输入的模型描述附件，再将描述连同对话内容交给 DeepSeek 主模型。自动模式会在可用时选择 Flash Vision Exp，同时继续支持显式配置其他 VS Code 模型或 API 端点。
+- **DeepSeek V4.1 Flash 和 Flash Vision Exp** 直接处理图片附件，不经过视觉代理。
+- **DeepSeek V4 Flash 和 Pro** 入口使用视觉代理：先由支持图片输入的模型描述附件，再将描述连同对话内容交给 DeepSeek 主模型。自动模式会在可用时选择 Flash Vision Exp，同时继续支持显式配置其他 VS Code 模型或 API 端点。
 
-如果你在意 DeepSeek 前缀缓存的复用，不建议只为查看一张图片而在对话中途切换模型。需要原生视觉时，可以从对话开始就选择 Flash Vision Exp；希望继续使用 Flash/Pro 时，则让视觉代理处理图片并保留主模型选择。
+[上下文缓存](https://api-docs.deepseek.com/zh-cn/guides/kv_cache/)命中需要完整匹配已有的缓存前缀。新对话推荐选择 V4.1 Flash，旧对话可按进度安排迁移。
 
 <p align="center">
   <img src="resources/screenshots/03-vision.png" alt="将图片拖入 Copilot Chat，DeepSeek 通过视觉代理响应" width="800">
 </p>
 
-### 思考模式与推理深度控制
-完整支持 DeepSeek V4 的 `reasoning_content`。Flash、Pro 和 Flash Vision Exp 均可选择 `停用`、`轻量`、`标准`（均衡，默认）或 `深度`（适用于复杂 Agent 任务），与官方 API 已实现的推理档位保持一致。
+### 思考模式与思考强度控制
+支持 DeepSeek 的思考模式和 `reasoning_content`。`停用`关闭思考模式；开启后可选择`轻量`（`low`）、`标准`（`high`，默认）或`深度`（`max`）。
 
 ### 继承全部 Copilot 能力
 由于本扩展接入的是 Copilot 的原生 provider API，你免费获得完整能力栈：
@@ -86,39 +86,41 @@ API Key 存储在 VS Code 的 `SecretStorage` 中（macOS 钥匙串 / Windows 
 
 1. 通过命令面板（`Cmd+Shift+P`）运行 **DeepSeek: 设置 API Key**
 2. 粘贴你的 Key 或兼容的 provider token（官方 DeepSeek Key 通常以 `sk-` 开头）
-3. 打开 Copilot Chat，点击模型选择器，选择 **DeepSeek V4 Flash**、**DeepSeek V4 Pro** 或 **DeepSeek V4 Flash Vision Exp**
+3. 打开 Copilot Chat，点击模型选择器，选择 **DeepSeek V4.1 Flash**
 4. 搞定——开始聊天
 
 ## 模型
 
-| 模型 | 图片处理 | 思考深度 | 适用场景 |
+| 模型入口 | 图片处理 | 思考强度 | 官方 API 状态 |
 |---|---|---|---|
-| **DeepSeek V4 Flash** | 视觉代理 | `停用` / `轻量` / `标准` / `深度` | 日常快速编码、小改动、低成本迭代 |
-| **DeepSeek V4 Pro** | 视觉代理 | `停用` / `轻量` / `标准` / `深度` | 复杂重构、Agent 任务、深度推理 |
-| **DeepSeek V4 Flash Vision Exp** | 原生图片输入 | `停用` / `轻量` / `标准` / `深度` | 直接、实验性的图片理解与快速推理 |
+| **DeepSeek V4.1 Flash** | 原生图片输入 | `停用` / `轻量` / `标准` / `深度` | 新对话推荐使用 |
+| **DeepSeek V4 Flash** | 视觉代理 | `停用` / `轻量` / `标准` / `深度` | 原模型已下线；旧模型名仍可调用 |
+| **DeepSeek V4 Pro** | 视觉代理 | `停用` / `轻量` / `标准` / `深度` | 将于北京时间 2026 年 9 月 14 日 12:00 下线 |
+| **DeepSeek V4 Flash Vision Exp** | 原生图片输入 | `停用` / `轻量` / `标准` / `深度` | 原模型已下线；旧模型名仍可调用 |
 
-三者均支持可选的思考模式、工具调用和 1M Token 上下文。Flash Vision Exp 仍是实验模型；直接使用时，自定义 API 端点或兼容服务商必须提供为它配置的模型 ID。
+四个入口均支持思考模式、工具调用和 1M Token 上下文。官方 API 的下线、路由和计费说明见[模型与价格文档](https://api-docs.deepseek.com/zh-cn/quick_start/pricing/)。
 
 ## 设置项
 
 | 设置项 | 默认值 | 说明 |
 |---|---|---|
 | `deepseek-copilot.baseUrl` | `https://api.deepseek.com` | API 端点——可改为自托管或代理部署地址 |
 | `deepseek-copilot.requestHeaders` | `{}` | 聊天补全请求的自定义请求头。[配置说明](https://github.com/Vizards/deepseek-v4-for-copilot/blob/main/docs/settings/request-headers.zh.md) |
-| `deepseek-copilot.maxTokens` | `0` | 最大输出 Token 数（`0` = 不限制）。可用于成本控制 |
-| `deepseek-copilot.modelIdOverrides` | 预填官方 ID 映射 | DeepSeek V4 Flash、Pro 和 Flash Vision Exp 对应的 API 模型 ID。仅在使用模型名不同的兼容第三方 API 时修改 |
+| `deepseek-copilot.maxTokens` | `0` | 最大输出 Token 数（`0` = API 默认值）。可用于成本控制 |
+| `deepseek-copilot.modelIdOverrides` | 预填官方 ID 映射 | V4.1 Flash 和三个旧 V4 入口对应的 API 模型 ID。仅在使用模型名不同的兼容第三方 API 时修改 |
 | `deepseek-copilot.debugMode` | `minimal` | 诊断模式：`minimal` 仅上报 token 用量，`metadata` 输出隐私安全日志，`verbose` 将完整请求 dump 和 pipeline snapshot 写入扩展 global storage。完整 dump 可能包含敏感提示词文本、工具定义、文件片段和图片描述。使用 `DeepSeek: 打开请求 Dump 目录` 打开 dump 位置 |
-| `deepseek-copilot.visionModel` | *(自动)* | Flash 和 Pro 使用的视觉代理。自动模式会在可用时选择 Flash Vision Exp；也可通过 `DeepSeek: 配置视觉代理` 改用其他 VS Code 模型或 API 端点 |
-| `deepseek-copilot.visionPrompt` | *(内置)* | Flash/Pro 的视觉代理用于描述图片附件的提示词，不影响 Flash Vision Exp 的原生图片请求 |
+| `deepseek-copilot.visionModel` | *(自动)* | V4 Flash 和 Pro 使用的视觉代理。自动模式会在可用时选择 Flash Vision Exp；也可通过 `DeepSeek: 配置视觉代理` 改用其他 VS Code 模型或 API 端点 |
+| `deepseek-copilot.visionPrompt` | *(内置)* | V4 Flash/Pro 的视觉代理用于描述图片附件的提示词，不影响 V4.1 Flash 或 Flash Vision Exp 的原生图片请求 |
 | `deepseek-copilot.experimental.stabilizeToolList` | `false` | 实验性设置。尝试预先激活 VS Code/Copilot 的虚拟工具，让传给 DeepSeek API 的 `tools` 参数在多轮对话中更完整、更稳定。当已启用工具跨轮次变化时，可能提高上下文缓存命中率。代价是 input tokens 可能增加；缓存命中的 input tokens 单价更低，但仍会计入用量。64 个或更少已启用工具时通常无需开启，除非工具列表仍在跨轮次变化；超过 128 个已启用工具时不建议开启 |
 
-思考深度可通过 Copilot Chat 的模型选择器对每个 DeepSeek 模型单独设置。
+思考强度可通过 Copilot Chat 的模型选择器对每个 DeepSeek 模型单独设置。
 
 兼容 API 代理的 `settings.json` 配置示例：
 
 ```json
 {
   "deepseek-copilot.modelIdOverrides": {
+    "deepseek-flash": "your-v4.1-flash-model-id",
     "deepseek-v4-flash": "your-flash-model-id",
     "deepseek-v4-pro": "your-pro-model-id",
     "deepseek-v4-flash-vision-exp": "your-vision-model-id"