🪶 Cline Copilot Chat: BYOK 33+ AI Models

Use 33+ AI models (DeepSeek V4, Kimi K2.7, GLM 5.2, GPT-5, Gemini 2.5, Grok 4, Qwen3.7, MiMo V2.5, MiniMax M3, Mistral, Llama, Sonar) in GitHub Copilot Chat. BYOK.

Bring Your Own Key (BYOK) · Cline (pay-per-use + free model) or ClinePass ($9.99/mo, $4.99 first month) · Works with native Copilot Agent Mode

✨ Why bother · ⚡ Quick Start (60 sec) · 🧠 Models · 📊 Compare · 🔧 Settings · ❓ FAQ · 💬 Community

---

💡 The pitch

Copilot Pro+ is $39 a month. The free tier caps you at 2,000 completions and a couple of models.

This extension adds Cline's models to the Copilot Chat picker you already use. Two options, same API key. Cline is pay-per-use across 23 models from 12 providers (GPT, Gemini, Grok, Claude, DeepSeek, Qwen, Kimi, GLM, MiMo, MiniMax, Mistral, Llama, plus Sonar and Command R+). One of them, DeepSeek V4 Flash, costs nothing at $0 balance. ClinePass is a flat $9.99/mo subscription ($4.99 first month) for 10 open-weight models with 2 to 5 times the standard rate limits.

You keep the native Copilot UI, tool-calling, Agent Mode. You just get more models to pick from, and the bill often comes out lower than Pro+.

---

🔥 Why bother

Copilot's model picker is locked to whatever GitHub decides to offer. This extension opens it up.

	What you get
💸 Cost	$0 with the free model. Or $9.99/mo ClinePass ($4.99 first month). Pay-per-use for anything in between.
🌍 Models	33 across 12 providers: DeepSeek V4, Kimi K2.7, GLM 5.2, Qwen3.7 Max, MiMo V2.5, MiniMax M3, GPT-5, Gemini 2.5, Grok 4, Mistral Large, Llama 4, Sonar Pro
🤖 Agent Mode	Tool-calling works: read files, edit, run terminal. Not just chat.
🧠 Thinking controls	Per-model reasoning effort. DeepSeek goes to max, Qwen takes a thinking_budget, MiniMax toggles, MiMo picks low/med/high.
🔌 Two providers, one key	Cline (pay-per-use) and ClinePass (subscription). Both active at once, switch from the picker.
🆓 Free model	DeepSeek V4 Flash returns 200 OK at $0 balance. No card needed.
🔒 Key storage	VS Code SecretStorage. The key stays on your machine.

---

⚡ Quick Start (60 sec)

1.  Install GitHub Copilot Chat (free) ──────────────────────────── ✓
2.  Install this extension ──────────────────────────────────────── ✓
3.  Get API key → app.cline.bot → Settings → API Keys ──────────── ✓
4.  Open Copilot Chat → model picker → "Add Models" → Cline ────── ✓
5.  Paste API key → pick a model → CHAT 🎉

📖 Detailed step-by-step

1. Install GitHub Copilot Chat first. Free, only needs a GitHub account.
2. Install this extension from the VS Code Marketplace. Or press F5 in this repo for dev mode.
3. Get an API key at app.cline.bot → Settings → API Keys.
   • DeepSeek V4 Flash works at $0 balance. Good for testing.
   • Want ClinePass instead? Subscribe at app.cline.bot/dashboard/subscription. $4.99 first month, then $9.99/mo.
4. Open Copilot Chat (Cmd/Ctrl+Shift+I, or click the Copilot icon).
5. Click the model picker (the current model name) → Add Models…
6. Pick Cline or ClinePass.
7. Press Enter for the default group name.
8. Paste your API key when asked. Stored in VS Code SecretStorage.
9. Pick a model. Start chatting. 🚀

💡 Tips:

• Cline and ClinePass are separate provider groups. Both can be active at once. Switch from the picker.
• One API key covers both.
• Model shows in Language Models but not the chat picker? Hover its row and click the eye icon (👁) to enable it.

---

📊 GitHub Copilot vs This Extension

GitHub Copilot has four tiers: Free, Pro ($10/mo), Pro+ ($39/mo), and Max ($100/mo). This is how BYOK via Cline stacks up:

	Copilot Free	Copilot Pro $10/mo	Copilot Pro+ $39/mo	Cline for Copilot Chat
💰 Cost	$0	$10/mo	$39/mo	$0 with free model. ClinePass $9.99/mo ($4.99 first month).
🤖 Models	GPT-5 mini, Haiku 4.5 (2,000 completions)	Pro catalog + Claude Code/Codex agents	Premium (Opus)	33 models: DeepSeek V4, Kimi K2.7, GLM 5.2, Qwen3.7, MiMo V2.5, MiniMax M3, GPT-5, Gemini 2.5, Grok 4, plus a free one
🧠 Reasoning controls	None	Per-model (GitHub decides)	Per-model (GitHub decides)	Per-family thinking effort you control
🔧 Agent Mode / tool-calling	None	Yes	Yes	Yes. Read, edit, terminal.
🎁 Free model?	No	No	No (paid tier only)	Yes. DeepSeek V4 Flash at $0 balance.
🚫 Rate limit	2,000 completions/mo	Unlimited (rate-limited)	4× Pro credits	Pay-per-use, or ClinePass 2-5× limits
🔌 Provider	GitHub only	GitHub only	GitHub only	Bring any Cline key

Not a replacement. This extension extends Copilot Chat. You still need the free Copilot Chat extension and a GitHub account. BYOK models bypass Copilot billing entirely. You pay Cline directly, or nothing at all on the free model.

---

🧠 Models

Cline: Pay-Per-Use (23 models)

Models are billed per token. No subscription required, just an API key and credits.

Model	ID	Context	Max Output
DeepSeek V4 Flash ⭐	deepseek/deepseek-v4-flash	1M	384K
DeepSeek V4 Pro	deepseek/deepseek-v4-pro	1M	384K
DeepSeek V3	deepseek/deepseek-v3	64K	8K
DeepSeek R1	deepseek/deepseek-r1	64K	16K
DeepSeek Chat	deepseek/deepseek-chat	64K	8K
GPT-4o	openai/gpt-4o	128K	16K
GPT-5	openai/gpt-5	256K	16K
o3	openai/o3	200K	100K
Gemini 2.5 Pro	google/gemini-2.5-pro	1M	65K
Grok 3	xai/grok-3	131K	16K
Grok 4	xai/grok-4	256K	16K
GLM 5.2	zai/glm-5.2	1M	128K
Kimi K3	moonshot/kimi-k3	1M	131K
Kimi K2.7 Code	moonshot/kimi-k2.7-code	262K	262K
Kimi K2.6	moonshot/kimi-k2.6	262K	65K
MiMo V2.5	mimo/mimo-v2.5	1M	128K
MiMo V2.5 Pro	mimo/mimo-v2.5-pro	1M	128K
MiniMax M3	minimax/minimax-m3	192K	131K
Qwen3.8 Max	qwen/qwen3.8-max	1M	65K
Qwen3.7 Max	qwen/qwen3.7-max	1M	65K
Qwen3.7 Plus	qwen/qwen3.7-plus	1M	65K
Mistral Large	mistral/mistral-large	128K	8K
Llama 4 Maverick	meta/llama-4-maverick	1M	8K
Sonar Pro	perplexity/sonar-pro	127K	8K
Command R+	cohere/command-r-plus	128K	4K

⭐ Free model. deepseek/deepseek-v4-flash returns 200 OK even at $0 balance.

All model IDs validated directly against the Cline API. anthropic/claude-* models are NOT currently available on Cline's API despite being listed in their docs.

ClinePass: $9.99/mo Subscription (12 models)

Open-weight models with 2 to 5× rate limits vs direct API access. No per-token charges.

Model	ID	Context	Max Output	Vision	Reasoning
DeepSeek V4 Flash	cline-pass/deepseek-v4-flash	1M	384K	❌	✅
DeepSeek V4 Pro	cline-pass/deepseek-v4-pro	1M	384K	❌	✅
GLM 5.2	cline-pass/glm-5.2	1M	128K	❌	✅
Kimi K3	cline-pass/kimi-k3	1M	131K	✅	✅
Kimi K2.7 Code	cline-pass/kimi-k2.7-code	262K	262K	✅	✅
Kimi K2.6	cline-pass/kimi-k2.6	262K	65K	✅	✅
MiMo V2.5	cline-pass/mimo-v2.5	1M	128K	❌	✅
MiMo V2.5 Pro	cline-pass/mimo-v2.5-pro	1M	128K	❌	✅
MiniMax M3	cline-pass/minimax-m3	192K	131K	✅	✅
Qwen3.8 Max	cline-pass/qwen3.8-max	1M	65K	✅	✅
Qwen3.7 Max	cline-pass/qwen3.7-max	1M	65K	❌	✅
Qwen3.7 Plus	cline-pass/qwen3.7-plus	1M	65K	✅	✅

Subscribe at app.cline.bot/dashboard/subscription

ClinePass Reference Pricing

You pay a flat $9.99/mo. These per-token rates are for quota comparison only:

Model	Input	Output	Cached Read
DeepSeek V4 Flash	$0.14/M	$0.28/M	$0.003/M
DeepSeek V4 Pro	$1.74/M	$3.48/M	$0.015/M
MiMo V2.5	$0.14/M	$0.28/M	$0.003/M
MiMo V2.5 Pro	$1.74/M	$3.48/M	$0.015/M
MiniMax M3	$0.30/M	$1.20/M	$0.06/M
GLM 5.2	$1.40/M	$4.40/M	$0.26/M
Kimi K2.7 Code	$0.95/M	$4.00/M	$0.19/M
Kimi K2.6	$0.95/M	$4.00/M	$0.16/M
Qwen3.7 Plus	$0.40/M	$1.60/M	$0.04/M
Qwen3.7 Max	$2.50/M	$7.50/M	$0.50/M

---

🔧 Settings

Setting	Default	Description
clineCopilotChat.temperature	0.2	Sampling temperature
clineCopilotChat.maxTokens	0	Max output tokens (0 = model default)
clineCopilotChat.requestTimeoutSeconds	600	Request timeout
clineCopilotChat.streamIdleTimeoutSeconds	120	Stream idle timeout
clineCopilotChat.debugReasoning	false	Log reasoning content to output channel
clineCopilotChat.stripThinkTags	auto	Handle <think> tags in output
clineCopilotChat.thinking.*	off	Per-family thinking mode (deepseek, glm, kimi, minimax, mimo, qwen)

---

❓ FAQ

How is this different from the Cline extension? Cline is a full autonomous coding agent. File edits, terminal, browser, the works. This extension does something narrower: it puts Cline's models into the Copilot Chat picker you already have. Same Chat, same Agent Mode, more models to choose from.

Cline vs ClinePass? Same API key, same endpoint. Different billing. Cline is pay-per-use across 23 models (GPT, Gemini, Grok, and friends). ClinePass is $9.99/mo flat for 10 open-weight models with 2 to 5 times the rate limits.

Can I use this without paying? Yes. DeepSeek V4 Flash returns 200 OK at $0 balance. Create a key at app.cline.bot and you're in.

Does Agent Mode work? Yes. All models support tool calling. Both Chat and Agent Mode.

Why not just use Copilot's built-in models? Copilot Free gives you a couple. This adds 33 from 12 providers, with a free one included. Pay-per-use or flat subscription, your call.

Rate limits? ClinePass gives 2-5× the standard API limits. Measured over 5-hour rolling, weekly, and monthly windows. Check usage at app.cline.bot.

---

� Troubleshooting

HTTP 403 — "only available via Cline product surfaces"

Some models were moved by Cline off the public API-key path and now only work inside the Cline app (IDE extension / CLI) with an account sign-in. Requesting them through an API key returns:

Error 403: <model> is only available via Cline product surfaces.

Your API key is fine — the model itself is what's locked out of the API.

What the extension does: confirmed-blocked models are hidden from the picker, and the error message shown in chat explains the cause instead of a generic 403. Confirmed so far: deepseek/deepseek-v4-flash (verified Aug 2026, issue #3).

Fix: pick a model that is still served over the API — e.g. deepseek/deepseek-chat, minimax/minimax-m3, or any ClinePass model — or use the Cline IDE extension / CLI for the blocked model with account sign-in.

To check whether a specific model still works with your key:

curl -X POST https://api.cline.bot/api/v1/chat/completions \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"deepseek/deepseek-chat","messages":[{"role":"user","content":"hi"}],"stream":false}'

• 200 — the model works via API key
• 403 ... product surfaces — blocked; report it in issue #3 so it can be added to the hidden list
• 401 — your key needs re-creating at app.cline.bot

Models missing from the picker on a fresh install or a second machine

VS Code Settings Sync does not sync SecretStorage for security reasons. When you install the extension on a new machine and sign in with Settings Sync, the extension downloads, but your API key does not — so provideLanguageModelChatInformation returns an empty list and the picker shows zero Cline / ClinePass models.

Fix:

1. Run Cline Copilot Chat: Set API Key from the Command Palette.
2. Paste your Cline API key (from app.cline.bot → Settings → API Keys).
3. Run Developer: Reload Window.

A warning toast with a Set API Key button now appears automatically the first time the extension activates without a key — you can skip step 1 and just click the button.

To verify the fix worked, open the Output panel (Cmd+Shift+U) and select the Cline Copilot Chat channel. You should see an activation banner ending with:

[activate] selectChatModels({ vendor: "cline" }): 13 model(s) visible to VS Code
[activate] selectChatModels({ vendor: "cline-pass" }): 10 model(s) visible to VS Code

The gear icon / "Manage Models…" does nothing when clicked

The gear icon invokes VS Code's built-in Manage Language Models command, whose precondition requires either an active Copilot entitlement or the github.copilot.clientByokEnabled context key. Since v0.1.4 the extension forces that context key to true when at least one model is registered, which keeps the gear clickable in most cases.

If it's still unresponsive after a window reload:

1. Run Developer: Reload Window — the context key is set on activation.
2. If still dead, sign in to GitHub Copilot Chat (a free personal GitHub account is enough — no Copilot Pro subscription required for BYOK). Signing in sets chatIsEnabled = true, which is the definitive fix.

Diagnosing "models not showing up"

Open the Output panel and select Cline Copilot Chat. The activation banner tells you exactly where the pipeline broke:

Banner line	What it means	Fix
SecretStorage ... MISSING	No API key on this machine	Run Cline Copilot Chat: Set API Key, then reload
selectChatModels ... 0 model(s) while key is present	Vendor contribution removed from package.json, or dual-provider race (fixed in v0.1.4 — upgrade)	Reinstall v0.1.4+
selectChatModels ... N model(s) but picker still empty	Picker cache stale per window	Developer: Reload Window
setContext ... FAILED	Context service unavailable	Reload window; if persistent, sign in to Copilot Chat

---

�📁 Project Structure

src/
├── extension.ts      # LanguageModelChatProvider, Cline vendor
├── providerTypes.ts  # Vendor constant
├── metadata.ts       # Model limits & capabilities
├── thinking.ts       # Thinking mode for 6 model families
├── streaming.ts      # SSE chat-completions streaming
├── errors.ts         # Error handling
└── retry.ts          # Retry logic

---

🛠️ Development

npm install          # Install dependencies
npm run compile      # Compile TypeScript
npm run watch        # Watch mode
# Press F5 to launch Extension Development Host

---

📄 License

MIT. See LICENSE.

---

💬 Community

GitHub Issues · X / Twitter · Reddit

If this saved you money or got a model working you needed, ⭐ star the repo.

📝 Found this useful? Leave a review on the VS Code Marketplace. It helps others find the extension.

🚀 Also from this publisher

Looking for OpenCode Zen / Go models in Copilot Chat? Check out OpenCode for Copilot Chat. 5,000+ installs, 30+ models, rotating free tier.

---

Independent project. Not affiliated with GitHub, Microsoft, Cline Bot Inc., or any model provider.

⬆ Back to top