Vizards/deepseek-v4-for-copilot

Pick DeepSeek V4 from the Copilot Chat model picker — and keep everything else Copilot already gives you.

What it solves

It allows users to use DeepSeek models (like V4.1 Flash and V4 Pro) directly within the GitHub Copilot Chat interface in VS Code. This eliminates the need to switch to a separate chat UI or a standalone extension while retaining all of Copilot's native features like agent mode, tool calling, and workspace search.

How it works

The extension plugs into Copilot's native provider API, adding DeepSeek models to the existing model picker. It supports a "Bring Your Own Key" (BYOK) model where the user provides their own DeepSeek API key. For image handling, it offers native vision support for specific models (V4.1 Flash) and a "Vision Proxy" for others, where a separate image-capable model describes the image before passing the text description to the main model.

Who it’s for

Developers who use GitHub Copilot but prefer the price-performance or capabilities of DeepSeek models and want to maintain their existing Copilot workflow and toolset.

Highlights

  • Seamless Integration: Adds models to the native Copilot Chat picker without adding new sidebars or UIs.
  • Full Feature Parity: Supports Copilot's agent mode, tool calling, .instructions.md files, and MCP skills.
  • Flexible Vision: Combines native image input with a configurable Vision Proxy for models that don't natively support images.
  • Thinking Mode: Supports DeepSeek's reasoning capabilities with configurable effort levels (none, low, high, max).
  • Secure Storage: Stores API keys in the OS keychain via VS Code's SecretStorage rather than in plain text files.

Related

  • Project
    VicBilibily/GCMP

    A VS Code extension that integrates various domestic Chinese LLM providers and custom API-compatible models into GitHub Copilot Chat.

  • Dispatch
    GitHub Copilot Internal Architecture: Context Injection and Session Storage Analysis

    A technical deep dive into GitHub Copilot's network traffic and source code reveals how it handles context injection, model routing, and the plaintext storage of user prompts in a local SQLite database.

  • Dispatch
    DeepSeek-V4-Flash-Vision-Exp Release Notes

    DeepSeek has released DeepSeek-V4-Flash-Vision-Exp, an experimental multimodal model that bridges the gap between fast text processing and high-performance visual understanding for agentic workflows.

  • Project
    lessweb/deepcode-cli

    A terminal-based AI coding assistant specifically optimized for DeepSeek models, featuring agent skills, MCP integration, and granular permission controls.