copilot-custom-endpoint 1.1.0 → 1.2.0

package/README.md

@@ -29,6 +29,7 @@ This repo is for those situations: validated, copy-paste-ready configs when Open
 | **Xiaomi MiMo**               | `mimo-v2.5`     | No                                 | ✅         | ✅        | ✅           | ✅⁴    |
 | **Xiaomi MiMo**               | `mimo-v2.5-pro` | No                                 | ✅         | ✅        | ✅           | ❌     |
 | **Xiaomi MiMo**               | `mimo-v2-flash` | No                                 | ✅         | ✅        | ✅           | ❌     |
+| **MiniMax**                   | `MiniMax-M3`    | No                                 | ✅         | ✅        | ✅           | ✅     |
 
 ¹ Proxy is optional: direct path works with static `enable_thinking: false`. Proxy adds dynamic thinking suppression (thinking ON in plain chat, OFF in tool loops).  
 ² With proxy: reasoning visible in plain chat. Without proxy: always suppressed.  
@@ -156,10 +157,37 @@ Here's a complete, real-world example of `chatLanguageModels.json` combining all
         }
       }
     ]
+  },
+  {
+    "name": "MiniMax",
+    "vendor": "customendpoint",
+    "apiKey": "<your-minimax-api-key>",
+    "apiType": "chat-completions",
+    "models": [
+      {
+        "id": "MiniMax-M3",
+        "name": "MiniMax M3",
+        "url": "https://api.minimax.io/v1/chat/completions",
+        "toolCalling": true,
+        "vision": true,
+        "streaming": true,
+        "maxInputTokens": 1048576,
+        "maxOutputTokens": 131072,
+        "requestBody": {
+          "thinking": { "type": "adaptive" },
+          "reasoning_split": true,
+          "temperature": 1,
+          "top_p": 0.95
+        }
+      }
+    ]
   }
 ]
 ```
 
+<details>
+<summary>Kimi K2.6 (Moonshot)</summary>
+
 ### Kimi K2.6 (Moonshot)
 
 #### 1. Grab a Moonshot API key
@@ -272,33 +300,92 @@ Open (or create) your user config file (see [Config file location](#config-file-
 | `invalid temperature` / `invalid top_p` | You're talking directly to Moonshot instead of through the proxy. Double-check the `url` in `chatLanguageModels.json`.                                        |
 | Tool calls fail after first turn        | This happens if "thinking" stays enabled during tool loops. The proxy normally disables it automatically; ensure you're on the latest `proxy/kimi-proxy.mjs`. |
 
+</details>
+
 ---
 
+<details>
+<summary>Qwen 3.6 Plus / Qwen 3.7 Max (DashScope)</summary>
+
 ### Qwen 3.6 Plus or Qwen 3.7 Max (DashScope)
 
-These models work with the optional `proxy/qwen-proxy.mjs` for dynamic thinking suppression (reasoning visible in plain chat, suppressed in tool loops). They also work **without a proxy** using a static `enable_thinking: false` — see the [direct path alternative](#direct-path-no-proxy) below.
+Qwen models work **directly** with DashScope — no proxy needed. Just add `enable_thinking: false` to `requestBody` for tool-calling stability. An optional `proxy/qwen-proxy.mjs` is also available for dynamic thinking suppression (see [below](#optional-local-proxy-for-dynamic-thinking)).
 
 #### 1. Grab a DashScope API key
 
-Sign up at [dashscope.aliyun.com](https://dashscope.aliyun.com) and create an API key.
+Create an API key [here](https://modelstudio.console.alibabacloud.com/ap-southeast-1?tab=dashboard#/api-key).
 
-#### 2. Start the optional local proxy (recommended)
+> **Regional endpoints:** DashScope offers endpoints for several regions. API keys are region-specific.
+>
+> - **China (Beijing):** `https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions`
+> - **US (Virginia):** `https://dashscope-us.aliyuncs.com/compatible-mode/v1/chat/completions`
+> - **Singapore (default):** `https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions`
 
-The proxy dynamically enables thinking in plain chat and disables it during tool calls:
+#### 2. Register the models in VS Code
 
-Run Qwen proxy
+Open (or create) your user config file (see [Config file location](#config-file-location) above) and paste this entry (replace `<your-dashscope-key>`):
+
+```json
+{
+  "name": "Qwen",
+  "vendor": "customendpoint",
+  "apiKey": "<your-dashscope-key>",
+  "apiType": "chat-completions",
+  "models": [
+    {
+      "id": "qwen3.7-max",
+      "name": "Qwen 3.7 Max",
+      "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
+      "toolCalling": true,
+      "vision": false,
+      "streaming": true,
+      "requestBody": {
+        "enable_thinking": false
+      }
+    },
+    {
+      "id": "qwen3.6-plus",
+      "name": "Qwen 3.6 Plus",
+      "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
+      "toolCalling": true,
+      "vision": true,
+      "streaming": true,
+      "requestBody": {
+        "enable_thinking": false
+      }
+    }
+  ]
+}
+```
+
+> **Trade-off:** `enable_thinking: false` suppresses reasoning in all requests (both plain chat and tool loops). Tool loops stay stable, but you never see the model's thought process. The [optional proxy](#optional-local-proxy-for-dynamic-thinking) below avoids this trade-off.
+
+#### 3. Chat!
+
+- Open the Copilot chat panel (`Ctrl+Alt+I` / `Cmd+Ctrl+I`).
+- Click the model picker (top-right of the chat input).
+- Choose **Qwen 3.6 Plus** (with vision) or **Qwen 3.7 Max** (text only).
+- Ask something. Streaming, tool use, and vision (3.6 Plus) all work.
+
+---
+
+#### Optional: Local proxy for dynamic thinking
+
+If you want reasoning visible in plain chat but automatically suppressed during tool loops, run the optional `proxy/qwen-proxy.mjs` instead.
+
+Start the proxy:
 
 ```bash
 npm run proxy:qwen
 ```
 
-Run all proxies
+Or with all proxies:
 
 ```bash
 npm run proxy
 ```
 
-Run globally (from any directory)
+Or globally (from any directory):
 
 ```bash
 # Qwen only
@@ -307,14 +394,6 @@ npx copilot-custom-endpoint qwen
 npx copilot-custom-endpoint
 ```
 
-Clean up debug logs
-
-```bash
-npm run clean:logs
-# or with npx
-npx copilot-custom-endpoint clean
-```
-
 You should see:
 
 ```
@@ -338,9 +417,7 @@ Expected response:
 }
 ```
 
-#### 3. Register the models in VS Code
-
-Open (or create) your user config file (see [Config file location](#config-file-location) above) and paste this entry (replace `<your-dashscope-key>`). Point URLs at the proxy and omit `requestBody` — the proxy handles thinking dynamically:
+Then update your VS Code config to point URLs at the proxy and remove `requestBody` — the proxy handles thinking dynamically:
 
 ```json
 {
@@ -371,59 +448,7 @@ Open (or create) your user config file (see [Config file location](#config-file-
 
 > **Keep the proxy terminal open** while using these models.
 
-#### 4. Chat!
-
-- Open the Copilot chat panel (`Ctrl+Alt+I` / `Cmd+Ctrl+I`).
-- Click the model picker (top-right of the chat input).
-- Choose **Qwen 3.6 Plus** (with vision) or **Qwen 3.7 Max** (text only).
-- Ask something. Streaming, tool use, and vision (3.6 Plus) all work.
-
-> **Regional endpoints:** If connecting directly (no proxy), DashScope offers endpoints for several regions. The proxy uses `dashscope-intl.aliyuncs.com` (Singapore) by default, configurable via `QWEN_UPSTREAM_URL`.
->
-> - **China (Beijing):** `https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions`
-> - **US (Virginia):** `https://dashscope-us.aliyuncs.com/compatible-mode/v1/chat/completions`
-> - **Singapore:** `https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions` (proxy default)
->
-> API keys are region-specific.
-
-#### Direct path (no proxy)
-
-If you prefer not to run the proxy, Qwen models work **directly** with DashScope by using the upstream URL and a static `enable_thinking: false` in `requestBody`:
-
-```json
-{
-  "name": "Qwen",
-  "vendor": "customendpoint",
-  "apiKey": "<your-dashscope-key>",
-  "apiType": "chat-completions",
-  "models": [
-    {
-      "id": "qwen3.7-max",
-      "name": "Qwen 3.7 Max",
-      "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
-      "toolCalling": true,
-      "vision": false,
-      "streaming": true,
-      "requestBody": {
-        "enable_thinking": false
-      }
-    },
-    {
-      "id": "qwen3.6-plus",
-      "name": "Qwen 3.6 Plus",
-      "url": "https://dashscope-intl.aliyuncs.com/compatible-mode/v1/chat/completions",
-      "toolCalling": true,
-      "vision": true,
-      "streaming": true,
-      "requestBody": {
-        "enable_thinking": false
-      }
-    }
-  ]
-}
-```
-
-> **Trade-off:** `enable_thinking: false` suppresses reasoning in all requests (both plain chat and tool loops). Tool loops stay stable, but you never see the model's thought process. The proxy path avoids this trade-off.
+The proxy URL is configurable via the `QWEN_UPSTREAM_URL` environment variable (defaults to the Singapore endpoint shown in [step 1](#1-grab-a-dashscope-api-key)).
 
 #### Troubleshooting (Qwen)
 
@@ -432,8 +457,13 @@ If you prefer not to run the proxy, Qwen models work **directly** with DashScope
 | `reasoning_content` errors during tool loops | Ensure `enable_thinking: false` is present in `requestBody` for every Qwen model.       |
 | Vision images fail to upload                 | Use base64-encoded images; external image URLs may fail if DashScope cannot reach them. |
 
+</details>
+
 ---
 
+<details>
+<summary>DeepSeek V4 (VS Code Extension)</summary>
+
 ### DeepSeek V4 (VS Code Extension)
 
 DeepSeek V4 Pro & Flash are available via a **dedicated VS Code extension** rather than a raw custom endpoint. The extension plugs DeepSeek directly into Copilot Chat's model picker while preserving agent mode, tool calling, skills, and MCP support.
@@ -475,8 +505,13 @@ DeepSeek V4 is text-only, but the extension handles images automatically — dro
 
 > For the full official guide, see: [github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/github_copilot.md](https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/github_copilot.md)
 
+</details>
+
 ---
 
+<details>
+<summary>Xiaomi MiMo</summary>
+
 ### Xiaomi MiMo
 
 MiMo works **directly** — no proxy needed. Just add the provider entry to your VS Code config and select the model in the chat picker.
@@ -563,6 +598,80 @@ Open your user config file (see [Config file location](#config-file-location) ab
 | 400 error `reasoning_content` during tool loops | Ensure `thinking: { "type": "disabled" }` is present in `requestBody` for every MiMo model.                       |
 | Vision images fail to upload                    | Use `mimo-v2.5` (the only model with native vision). Text-only models (`pro`, `flash`) don't support image input. |
 
+</details>
+
+---
+
+<details>
+<summary>MiniMax M3 (MiniMax)</summary>
+
+### MiniMax M3 (MiniMax)
+
+MiniMax works **directly** with the OpenAI-compatible Chat Completions endpoint — no proxy needed. The recommended config enables MiniMax's native reasoning via `thinking: { "type": "adaptive" }` + `reasoning_split: true`.
+
+#### 1. Grab a MiniMax API key
+
+Create an API key at the [MiniMax Developer Platform](https://platform.minimax.io/user-center/basic-information/interface-key).
+
+> **Regional endpoints:** MiniMax offers endpoints for different regions. API keys are region-specific.
+>
+> - **International (default):** `https://api.minimax.io/v1/chat/completions`
+> - **China:** `https://api.minimaxi.com/v1/chat/completions`
+
+#### 2. Register the model in VS Code
+
+Open (or create) your user config file (see [Config file location](#config-file-location) above) and paste this entry (replace `<your-minimax-api-key>`):
+
+```json
+{
+  "name": "MiniMax",
+  "vendor": "customendpoint",
+  "apiKey": "<your-minimax-api-key>",
+  "apiType": "chat-completions",
+  "models": [
+    {
+      "id": "MiniMax-M3",
+      "name": "MiniMax M3",
+      "url": "https://api.minimax.io/v1/chat/completions",
+      "toolCalling": true,
+      "vision": true,
+      "streaming": true,
+      "maxInputTokens": 1048576,
+      "maxOutputTokens": 131072,
+      "requestBody": {
+        "thinking": { "type": "adaptive" },
+        "reasoning_split": true,
+        "temperature": 1,
+        "top_p": 0.95
+      }
+    }
+  ]
+}
+```
+
+**Why this config?**
+
+- `thinking: { "type": "adaptive" }` — MiniMax's documented default. The model decides when to reason.
+- `reasoning_split: true` — the server returns reasoning in a structured `reasoning_details` field instead of mixing `<think>` tags into `content`. VS Code sees a clean OpenAI-format message.
+
+> **Note:** `thinking: { "type": "disabled" }` is **not** a hard override — Phase 1 testing confirmed MiniMax-M3 still reasons internally regardless of this setting, and emits `<think>` tags in `content` either way. Setting it to `disabled` only changes the response field layout, not actual model behavior. We recommend `adaptive` for clarity.
+
+#### 3. Chat!
+
+- Open the Copilot chat panel (`Ctrl+Alt+I` / `Cmd+Ctrl+I`).
+- Click the model picker and select **MiniMax M3**.
+- Ask something. Plain chat, streaming, tool use, and vision all work.
+
+#### Troubleshooting (MiniMax)
+
+| Symptom                              | Fix                                                                                                           |
+| ------------------------------------ | ------------------------------------------------------------------------------------------------------------- |
+| Model not appearing in picker        | Check your `chatLanguageModels.json` syntax. Reload the VS Code window.                                       |
+| 400 on tool calls                    | Confirm the model ID is `MiniMax-M3` (capital M's, lowercase i, hyphen). Check the API key region.            |
+| Responses show leaked `<think>` tags | Make sure `"reasoning_split": true` is set in `requestBody` so reasoning goes to `reasoning_details` instead. |
+
+</details>
+
 ---
 
 For the full research notes, tested values, and known limitations, see:
@@ -570,12 +679,13 @@ For the full research notes, tested values, and known limitations, see:
 - [`docs/models/kimi-k2.6.md`](docs/models/kimi-k2.6.md)
 - [`docs/models/qwen.md`](docs/models/qwen.md)
 - [`docs/models/mimo.md`](docs/models/mimo.md)
+- [`docs/models/minimax.md`](docs/models/minimax.md)
 
 ## Pricing comparison
 
 > **⏰ June 1, 2026 — GitHub Copilot switched to usage-based billing (AI Credits) today.**
 >
-> Before this change, Copilot was a flat subscription — no per-turn metering, so you could use chat and agent mode as much as you wanted within rate-limit bounds. Now **every interaction burns AI credits** from your monthly allowance. Agent mode and complex multi-file tasks consume significantly more tokens than simple Q&A, which means your 7,000 Pro+ credits can disappear fast if you're using frontier models.
+> Before this change, Copilot used **premium request-based billing** — each model had its own multiplier (e.g., GPT-5.5 = 7.5×, Claude Sonnet 4.6 = 1×, Haiku 4.5 = 0.33×), and every request consumed `multiplier × 1` from your monthly premium-request allowance. Now **every interaction burns AI credits** based on actual token consumption. Agent mode and complex multi-file tasks consume significantly more tokens than simple Q&A, which means your 7,000 Pro+ credits can disappear fast if you're using frontier models.
 >
 > **The practical workaround:** use cheaper alternative models (DeepSeek V4 Flash, Kimi K2.6, Qwen) that are still powerful enough for coding — often at **5–55× less cost** than the Copilot defaults. The tables below show the exact comparison.
 >
@@ -625,6 +735,7 @@ These are the models available through GitHub Copilot's model roster as of June
 | **MiMo V2.5 Pro**     | Xiaomi    | $1.00                         | $3.00                                   | 1M             |
 | **Qwen 3.6 Plus**     | DashScope | $0.50 (≤256K) / $2.00 (>256K) | $3.00 (≤256K) / $6.00 (>256K)           | 1M             |
 | **Qwen 3.7 Max**      | DashScope | $2.50 (≤1M)                   | $7.50 (≤1M)                             | 1M             |
+| **MiniMax M3**        | MiniMax   | $0.60 (≤512K) / $1.20 (>512K) | $2.40 (≤512K) / $4.80 (>512K)           | 1M             |
 
 > **Notes:**
 >
@@ -636,6 +747,7 @@ These are the models available through GitHub Copilot's model roster as of June
 > - **Qwen** models use **tiered pricing** — determined by total input tokens per request. Prices above are for non-thinking mode.
 > - **Kimi K2.6** pricing is from the **Moonshot platform** (direct). Via DashScope: $0.89 input / $3.71 output.
 > - **DashScope** offers a **free quota** of 1M input + 1M output tokens per model, valid for 90 days.
+> - **MiniMax M3** uses **tiered pricing** — input price doubles above 512K input tokens. A 7-day 50% off promotion is available for new accounts.
 > - **MiMo** offers a **Token Plan** subscription model with discounted rates and a free cache-writing promotion.
 > - For typical Copilot chat usage (short-to-medium prompts), you'll almost always fall in the lowest pricing tier.
 
@@ -650,6 +762,7 @@ These are the models available through GitHub Copilot's model roster as of June
 | Kimi K2.6 (thinking)     | ~$0.48                 | —                    |
 | Gemini 3 Flash           | ~$0.55                 | ~55                  |
 | Qwen 3.6 Plus            | ~$0.55                 | —                    |
+| MiniMax M3               | ~$0.54                 | —                    |
 | MiMo V2.5 Pro            | ~$0.80                 | —                    |
 | GPT-5.4 mini             | ~$0.83                 | ~83                  |
 | Claude Haiku 4.5         | ~$1.00                 | ~100                 |
@@ -675,6 +788,7 @@ These are the models available through GitHub Copilot's model roster as of June
 > - [DashScope pricing](https://www.alibabacloud.com/help/en/model-studio/billing-for-model-studio)
 > - [DeepSeek pricing](https://api-docs.deepseek.com/quick_start/pricing)
 > - [MiMo pricing](https://platform.xiaomimimo.com/docs/en-US/pricing)
+> - [MiniMax pricing](https://platform.minimax.io/docs/pricing/overview)
 
 ## Repo layout
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "copilot-custom-endpoint",
-  "version": "1.1.0",
+  "version": "1.2.0",
   "description": "Local proxies for VS Code Copilot custom endpoints — Kimi K2 & Qwen 3.x",
   "license": "MIT",
   "type": "module",