💡 The elevator pitch

Copilot Chat is great — but its premium models cost $39/mo (Pro+), and the free tier is rate-limited. This extension plugs OpenCode's model gateway into Copilot Chat's model picker. OpenCode Zen gives you rotating free models (Big Pickle is currently free; DeepSeek V4 Flash, MiMo-V2.5, and others rotate) plus paid models like Claude Opus, GPT-5.5, and Gemini at pay-as-you-go rates. OpenCode Go ($10/mo, $5 first month promo) gives you a curated set of open models (DeepSeek V4 Pro, Kimi K3, GLM-5.2, Qwen3.8 Max, MiMo V2.5 Pro) with generous usage limits. You keep the native Copilot UI, tool-calling, and Agent Mode — you just get way more models, often cheaper than Copilot Pro+.

🔥 Why you'll love it

⚡ Quick Start (60 sec)

1.  Install or update VS Code 1.125+ ────────────────────────────── ✓
2.  Install this extension ──────────────────────────────────────── ✓
3.  Get an OpenCode Zen API key → opencode.ai/auth ─────────────── ✓
4.  Open Copilot Chat → click model → "Add Models" → OpenCode Zen ── ✓
5.  Paste API key → pick a free model → CHAT 🎉

🎬 Demo

Selecting an OpenCode model from the Copilot Chat model picker.

🧠 Models

The extension fetches live model lists on every startup from:

⭐ OpenCode Go (subscription — $10/mo, $5 first month promo)

Curated open coding models, refreshed live from the endpoint. Deprecated/legacy models (e.g. glm-5, kimi-k2.5, minimax-m2.5, mimo-v2-pro, mimo-v2-omni, qwen3.5-plus) are filtered before registration.

Usage: most Go models include ~$60/mo of usage; grok-4.5, gpt-5.6-luna, kimi-k3, qwen3.8-max, deepseek-v4-pro and mimo-v2.5-pro include ~$15/mo. If you hit a limit, keep using the free Zen models (or enable the "Use balance" option in the Zen console).

🆓 OpenCode Zen — Free models (no payment needed)

OpenCode Zen offers rotating free models — no balance required. Currently free: Big Pickle (stealth model), DeepSeek V4 Flash Free, MiMo-V2.5 Free, Hy3 Free, Laguna S 2.1 Free, Ling-3.0-tiny Free, Nemotron 3 Ultra Free, and Nemotron 3.5 Lightning Free. Without a balance, rate limits are low. Adding $20+ to your Zen balance significantly improves rate limits on free models.

Note: Per the OpenCode docs, free models are all offered for a limited time and rotate periodically. The table shows current offerings — availability may change.

💰 OpenCode Zen — Paid models (requires balance)

Add a payment method to your Zen account to unlock these models at pay-as-you-go rates. OpenCode passes provider price drops through at cost. Prices per 1M tokens.

Set opencodego.freeOnly: false to show paid Zen models in the picker (default shows only free models).

📊 GitHub Copilot vs This Extension

GitHub Copilot has four tiers now — Free, Pro ($10/mo), Pro+ ($39/mo), and Max ($100/mo). Here's how BYOK via OpenCode compares:

Model bridge, not completions replacement — OpenCode BYOK models work in VS Code Chat without a Copilot plan or GitHub sign-in. Standard inline suggestions, next-edit suggestions, semantic search, and embedding-backed features still require GitHub/Copilot support. OpenCode requests bypass Copilot billing entirely — you pay OpenCode directly (or nothing, on Zen free).

💡 When to use which?

• VS Code + OpenCode Zen → $0 total for BYOK chat. Best for students, hobbyists, and trying frontier models.
• Copilot Pro + OpenCode Go → $10/mo for Copilot's polish + $10/mo for DeepSeek Pro, Kimi K2.6, Qwen3.7 Max.
• This extension alone → Chat, agent tools, and configured utility tasks work without Copilot. Add Copilot only when you also need inline suggestions, semantic search, or embedding-backed features.

✨ Features Deep Dive

🧠 Thinking & Reasoning Controls

Per-model reasoning configuration, dynamically enhanced with reasoning_options from models.dev:

opencodego.debugReasoning — writes provider reasoning_content to Output → OpenCode for debugging.

Context Safety

• Request budgets include both conversation messages and tool/MCP schemas, with proportional headroom for differences between provider tokenizers.
• When an upstream provider returns exact context-overflow counts, the extension reduces the output budget and retries once automatically.

📊 Usage Tracking

• Go Usage Tracker — real-time burn-rate of OpenCode Go subscription:
  • Tracks 5-hour rolling ($12), weekly ($30), monthly ($60) tiers.
  • Server-synced meters — on startup and after each request the tracker pulls the official /zen/go/v1/usage endpoint with your Go key, so the Session/Weekly/Monthly percentages and "resets in" timers are account-wide and server-accurate (includes CLI, other devices, anything before the extension was installed). Falls back to local estimates when the endpoint is unreachable or the key/subscription is invalid.
  • Today / Yesterday and per-session spend stay device-local (the API does not return them).
  • Status bar: Go: 27%·62%·75% — ⚠ warning when any tier exceeds 80%.
  • Persisted in VS Code globalState — survives restarts.
• Response usage bar — latest prompt/output/total/cache summary after each response.
• Normalized usage DataPart — emits usage MIME so Copilot Chat's context widget shows accurate token counts.

Multiple Go accounts

You can use more than one OpenCode Go account in the same VS Code window. When a second Go API key is added via the Manage Language Models panel, the extension creates a new usage profile ("Profile 1", "Profile 2", etc.) on the first request made with each key.

• Auto-switch — the status bar and SVG hover card follow the model you last used. If you chat with a model from a different account, the panel switches automatically.
• QuickPick — click the status bar to see which profile is active and switch to another profile.
• Commands — OpenCode Go: Rename Active Profile and OpenCode Go: Delete Profile help you manage your profiles. The label you give a profile appears in the status bar and SVG title.
• Storage isolation — each profile has its own globalState storage namespace (opencodego.usageLog.v1.<fingerprint>) so data never mixes.
• SQLite — when 2+ profiles exist, the extension does not consult the shared opencode.db (which is per-machine and cannot be attributed to a specific API key). Accuracy falls back to in-memory entries for multi-account setups.

Upgrading from a single-account install? Your existing usage data is migrated into Profile 1 the first time a second profile is created. Your original stats are preserved — nothing is lost.

🪟 Agents Window (Copilot CLI) Support

OpenCode models appear in the VS Code Agents window model picker when starting a Copilot CLI / Background agent session — not just the regular Chat view:

How it works:

VS Code ≥1.129 runs the Agents window in a separate agent host process and keeps two knobs off by default that this extension depends on:

1. chat.agentHost.byokModels.enabled (experimental) — the BYOK language-model bridge that mirrors extension-provided BYOK models into agent-host sessions.
2. extensions.supportAgentsWindow — without it, code extensions are disabled in the Agents window process, so OpenCode Go/Zen don't appear in the Agents window picker nor in its + Add Models list.

The extension auto-enables both when agentsWindow is on:

Setup:

1. Agent models are enabled by default (agentsWindow: true). The extension auto-enables the required VS Code settings on first activation and offers a Reload Now button — a window reload is required for them to take effect.
2. Reload the window (Developer: Reload Window) if you haven't been prompted.
3. Open the Agents window → start a new session → select Copilot CLI as the agent type.
4. Open the model picker — OpenCode models appear under their provider; the + Add Models list in the Agents window's Language Models view now includes OpenCode Go and OpenCode Zen.

To manage agent API keys separately or see agent vendors in the Manage panel, enable:

"opencodego.showAgentModelsInManagePanel": true

❌ Removing a provider from Language Models

Like deleting a provider in GitHub Copilot's Manage Language Models, you can remove OpenCode Go or OpenCode Zen from the Language Models list and every model picker:

• Command Palette — OpenCode Go: Remove/Re-add Provider in Language Models (same for Zen), or use Manage Provider → Remove from Language Models.
• Settings — set opencodego.enabled / opencodezen.enabled to false.

The provider's vendor row and models disappear from the Manage Language Models view, the + Add Models list, the Chat picker, and the Agents window. Your API key and BYOK group settings are kept, so re-enabling (or the Re-add to Language Models action) restores everything. A window reload is required after toggling.

✍️ Inline Code Suggestions (Experimental)

Ghost-text completions while typing, powered by the OpenCode gateway with thinking forced off:

• Opt-in: "opencodego.inlineSuggestions": true (requires a window reload).
• Model: opencodego.inlineSuggestionsModel — defaults to qwen3.5-plus, whose enable_thinking=false mode is a genuine no-reasoning path (measured ~1.5s TTFB, zero hidden reasoning). Reasoning models (e.g. deepseek-v4-flash) burn 100+ reasoning tokens even with thinking off and are not recommended.
• Tuning knobs (all optional): inlineSuggestionsDebounceMs (300), inlineSuggestionsTimeoutMs (3000), inlineSuggestionsMaxTokens (128), inlineSuggestionsPrefixLines (10), inlineSuggestionsSuffixChars (300).
• Requests are tiny (10 lines before the cursor + a short suffix by default), debounced 300ms, time out at 3s, and are aborted on the next keystroke.
• The gateway exposes no FIM endpoint, so completions use fill-in-the-middle emulation with FIM tokens over /chat/completions.

🛠️ Smart Routing & Reliability

• Native endpoint routing per family (see Models table)
• Tool-calling forwarded in correct format per endpoint (OpenAI tool_calls vs Anthropic tool_use)
• Sticky gateway headers (x-opencode-session, x-opencode-request, x-opencode-client) for affinity
• Request & stream timeouts — defaults 600s total / 120s idle; configurable
• Transient 5xx retry — momentary gateway 502/503/504 and Router.Unavailable errors are retried up to 2× with exponential backoff + jitter
• ground tag filtering — opencodego.stripThinkTags (auto strips MiniMax only, always, never)

🔍 Diagnostics

Provider diagnostics also include the VS Code/extension versions, extension host, workspace trust, Windows elevation level, installation paths, credential presence, and model-selection errors.

🔧 Settings

🎛️ Commands

The easiest way to manage your key is Settings → Language Models (gear ⚙). For advanced use, open the Command Palette (Cmd/Ctrl+Shift+P):

❓ FAQ

🏗️ Architecture

flowchart LR
    A[VS Code Copilot Chat] -->|model picker| B{Which vendor?}
    B -->|Copilot| C[GitHub models]
    B -->|OpenCode Go| D[Go API Key]
    B -->|OpenCode Zen| E[Zen API Key]

    D --> F[Smart Router]
    E --> F

    F -->|GPT| G[" /responses "]
    F -->|Gemini| H[" :streamGenerateContent "]
    F -->|Claude + MiniMax| I[" /messages "]
    F -->|Qwen/DeepSeek/GLM/Kimi/MiMo| J[" /chat/completions "]

    G & H & I & J --> K[opencode.ai gateway]
    K --> L[SSE Stream]
    L -->|tool_calls / tool_use| A
    L -->|usage DataPart| M[Status Bar]
    L -->|context hook| N[Context Widget]

See docs/architecture/ for the full provider architecture, routing, and metadata resolution docs.

🤝 Contributing

Contributions welcome! Whether it's a typo fix, new model support, or a screenshot — every PR counts.

📋 See CONTRIBUTING.md for guidelines. 💬 Discussions: GitHub Discussions 🐞 Bugs: Issue Tracker

Development

npm install      # install deps
npm run compile  # build TypeScript
npm run watch    # watch mode
npm run package  # build .vsix

Press F5 in VS Code to launch an Extension Development Host.

📈 Roadmap

• [x] 🎥 Demo GIF + screenshots
• [x] 📦 Publish to VS Code Marketplace
• [x] 📊 Usage panel (status bar + webview)
• [ ] 🌐 GitHub Pages landing page
• [ ] 🏷️ Verified publisher badge
• [ ] 🔔 Webhook for new OpenCode models
• [ ] 🎨 Custom model aliases / favorites
• [ ] 🌍 i18n (id, zh, ja)

Have an idea? Start a discussion or open a feature request.

⭐ Star History

 👆 Star this repo if it saved you money or unlocked a model you needed!

ℹ️ Note: GitHub restricts the stargazers API (since June 30, 2026) to a repository's own admins and collaborators. The chart above only renders after you add a GitHub Access Token on star-history.com — use a fine-grained token with Metadata → Read-only + Contents → Read and write, or a classic token with the public_repo scope.

💬 Community

If this saves you money or unlocks a model you needed — ⭐ star the repo and share it!

📄 License

MIT © 2026 ltmoerdani

OpenCode is a trademark of opencode.ai. This project is independent and not affiliated with GitHub, Microsoft, Anthropic, OpenAI, Google, or any model provider.