# Msty Studio + QCode on Windows: GPT and Claude, step by step


**Short answer**: In Msty Studio 3.0.1, open **Settings → Model Providers → + Add Provider**, choose **Open AI Compatible**, paste your `cr_` API key, set **Inference Endpoint** to `https://api.qcode.cc/openai/v1`, click **Fetch Models**, check `gpt-6.2-sol`, name the provider `QCode`, and click **Save**. For Claude models, add a separate built-in **Claude** provider with **Inference Endpoint** set to `https://api.qcode.cc/api/v1`.

| Item | Fill in |
|---|---|
| Provider | `Open AI Compatible` |
| Inference Endpoint | `https://api.qcode.cc/openai/v1` |
| API Key | `cr_xxxxxxxx` |
| Model | `gpt-6.2-sol` |
| Built-in Claude endpoint | `https://api.qcode.cc/api/v1` |

Checked on 2026-11-22 · Msty Studio 3.0.1 · Windows 11


Follow the 10 screenshots below on Windows 11 to install Msty Studio 3.0.1, connect QCode through the **Open AI Compatible** provider, select `gpt-6.2-sol`, and receive your first reply. You do not need a Msty account or an Aurum license, and you can also add the built-in **Claude** provider at the end of this guide.


- **Applies to**: Windows 10 / 11 (64-bit) (Windows 11 English UI)
- **Software version**: Msty Studio 3.0.1
- **Screenshots updated**: 2026-11
- **Time needed**: about 10 minutes (includes downloading a ~228 MB installer)
- **You will need**: a QCode API key starting with `cr_`


Before the first screenshot, confirm that you have these ready:


- A 64-bit Windows 10 or Windows 11 PC (Intel or AMD for `x64`, or Windows 11 `ARM64` for ARM devices) that can open `msty.ai` and reach `api.qcode.cc`
- Roughly 1.1 GB of available disk space (~228 MB for the installer plus ~859 MB for the app in your user folder — no administrator rights are needed)
- No Msty account or Aurum license is required — both **Open AI Compatible** and the built-in **Claude** provider work out of the box
- A QCode API key copied from your dashboard (`cr_xxxxxxxx`). No account yet? [Sign up](https://qcode.cc/en/auth/sign-up?utm_source=docs.qcode.cc&utm_medium=tut_before&utm_campaign=knowledge_base&src=docs-ide-msty-before) first, pick a plan on the [pricing page](https://qcode.cc/en/pricing?utm_source=docs.qcode.cc&utm_medium=tut_before&utm_campaign=knowledge_base&src=docs-ide-msty-before), then create the key in your [Dashboard](https://qcode.cc/en/dashboard).


## Download and install Msty Studio

Msty Studio is distributed only through its official website and installer CDN (there are no GitHub Releases, winget packages, or Microsoft Store builds). Pick the tab that matches your region:

**Other regions**

Open [msty.ai/products/studio/](https://msty.ai/products/studio/) and click **Download Studio** (which auto-detects Windows x64), or open **Choose installer** to pick **Windows (x64)** (`Windows 10/11 · Intel or AMD`, `MstyStudio_x64.exe`, ~228 MB) or **Windows (ARM64)** (`Windows 11 · ARM64`, `MstyStudio_arm64.exe`).


**Mainland China**

Open the official download page [msty.ai/products/studio/](https://msty.ai/products/studio/) in your browser. Click **Choose installer** next to **Download Studio** and select **Windows (x64)** for Intel/AMD PCs (`MstyStudio_x64.exe`, ~228 MB) or **Windows (ARM64)** for ARM devices (`MstyStudio_arm64.exe`).

Installers are served from `next-assets.msty.studio` and there is no official regional mirror. If the download slows down or stalls, wait a few minutes and retry or switch networks rather than downloading from unofficial sites.




### Download the Windows installer

Open [msty.ai/products/studio/](https://msty.ai/products/studio/) and click **Choose installer** ① next to **Download Studio** to open the installer menu. Click **Windows (x64)** ② (`Windows 10/11 · Intel or AMD`) to download `MstyStudio_x64.exe` (pick **Windows (ARM64)** if your PC uses an ARM processor).

![Open the official Msty Studio download page, open the “Choose installer” menu, and select “Windows (x64)” to download MstyStudio_x64.exe](https://docs.qcode.cc/static/images/tut/msty/202611/en/01-download-page.webp)


Done right you will see: your browser starts downloading `MstyStudio_x64.exe` (~228 MB) into your **Downloads** folder.




### Install Msty Studio

Double-click `MstyStudio_x64.exe` in **Downloads**. On the first screen (**Choose Installation Options**), leave **Only for me** selected and click **Next >**. On **Choose Install Location**, keep the default folder under your user profile and click **Install** ①, then leave **Run MstyStudio** checked and click **Finish**.

![Run MstyStudio_x64.exe, keep “Only for me” and click Next, then click “Install” on the install location step (no UAC prompt required)](https://docs.qcode.cc/static/images/tut/msty/202611/en/02-installer.webp)

**Stuck? Windows SmartScreen shows a warning**

If Windows shows **Windows protected your PC**, click **More info** and then **Run anyway** to launch the official installer. Under the default **Only for me** option, no administrator prompt is needed.



Done right you will see: installation finishes without an administrator prompt, and Msty Studio opens automatically after you click **Finish**.




### Skip the first-launch onboarding

On first launch, Msty Studio asks **What brings you to Msty Studio?** and confirms **No account needed.** You can click **Just exploring** ① if you want to walk through the quick setup, or click **Skip to app →** ② in the lower right to go straight to the main window.

![First launch requires no account; pick “Just exploring” or click “Skip to app →” in the lower right to enter Msty Studio directly](https://docs.qcode.cc/static/images/tut/msty/202611/en/03-onboarding.webp)


Done right you will see: the main Msty Studio chat workspace opens without asking you to sign in or activate a license.




### Open Model Providers in Settings

Click **Settings** at the very bottom of the left sidebar. Under **PROVIDERS & ENGINES**, click **Model Providers** ①, then click the blue **+ Add Provider** ② button in the center of the page.

![Open Settings at the bottom left, select “Model Providers” in the sidebar, and click “+ Add Provider”](https://docs.qcode.cc/static/images/tut/msty/202611/en/04-model-providers.webp)

**Stuck? I cannot find “Model Hub > Model Providers”**

Msty Studio 3.x moved provider management out of **Model Hub** into **Settings → Model Providers** (under **PROVIDERS & ENGINES**). If you already added another provider earlier, **+ Add Provider** moves from the center to the top-right corner of the same page.



Done right you will see: a **New Model Provider** drawer slides open on the right side of the window.




### Choose Open AI Compatible

In the **New Model Provider** drawer, click the **Provider** dropdown (it defaults to **OpenAI**), type `compatible` into the search box ①, and select **Open AI Compatible** ② (notice the space between `Open` and `AI`).

![In the “New Model Provider” drawer, open the Provider dropdown, search for “compatible”, and choose “Open AI Compatible”](https://docs.qcode.cc/static/images/tut/msty/202611/en/05-select-provider.webp)

**Stuck? Why not keep the default “OpenAI” provider?**

The standard **OpenAI** option targets OpenAI's official host, whereas **Open AI Compatible** lets you enter QCode's `Inference Endpoint` directly and does not require an Aurum license. To use Claude models instead, see **Optional: use Claude models with the built-in Claude provider** below.



Done right you will see: the drawer updates to show **API Key**, **Inference Endpoint (must start with http:// or https://)**, and a **Fetch Models** button.




### Enter your API key and Inference Endpoint

Paste your QCode key (`cr_xxxxxxxx`) into **API Key** ① — the field masks the characters automatically. Enter `https://api.qcode.cc/openai/v1` into **Inference Endpoint (must start with http:// or https://)** ②, leaving out any `/chat/completions` suffix, then click **Fetch Models** ③.

![Paste your QCode API key, set Inference Endpoint to https://api.qcode.cc/openai/v1, and click “Fetch Models”](https://docs.qcode.cc/static/images/tut/msty/202611/en/06-endpoint-apikey.webp)

**Need an API Key?**

If you do not have a QCode account yet, [Sign up](https://qcode.cc/en/auth/sign-up?utm_source=docs.qcode.cc&utm_medium=tut_key_step&utm_campaign=knowledge_base&src=docs-ide-msty-key) first, buy a pack on the [Pay-as-you-go](https://qcode.cc/en/pay-as-you-go?utm_source=docs.qcode.cc&utm_medium=tut_key_step&utm_campaign=knowledge_base&src=docs-ide-msty-key) page, and then create an API key in your [Dashboard](https://qcode.cc/en/dashboard). Copy the full string starting with `cr_`. Credit and debit cards are accepted — see [Payment methods](/docs/reference/payment-methods) for details.



Done right you will see: **API Key** ① shows masked dots, **Inference Endpoint** ② reads `https://api.qcode.cc/openai/v1`, and clicking **Fetch Models** ③ loads a list of QCode models below the button.




### Select a model, name the provider, and save

Scroll down to the fetched **Models** list and check `gpt-6.2-sol` ① (you can check additional GPT or Chinese models if you like). In **Give it a unique name (Optional)** ② at the bottom of the drawer, type `QCode`, then click **Save** ③.

![After models are fetched, check gpt-6.2-sol, name the provider QCode at the bottom, and click “Save”](https://docs.qcode.cc/static/images/tut/msty/202611/en/07-select-model-save.webp)

**Stuck? The model you want is not in the list**

Click **+ Add Custom Model** directly above the model search box and enter the exact model ID (for example `gpt-6.2-sol`). Checking a model's box here is what makes it appear in the chat model selector — there is no separate “Show in Model Selector” switch in Msty Studio 3.0.1.



Done right you will see: `gpt-6.2-sol` ① is checked, `QCode` ② appears in the bottom name field, and clicking **Save** ③ closes the drawer.




### Confirm the saved provider and return to chat

Back on **Model Providers**, confirm that **QCode · Open AI Compatible** ① appears with `1 Model` (or the count you checked) and its toggle switch on the right is enabled. Click **← Back** ② in the top-left corner to return to the chat screen.

![“QCode · Open AI Compatible (1 Model)” now appears enabled in Model Providers; click “← Back” at the top left to return to chat](https://docs.qcode.cc/static/images/tut/msty/202611/en/08-provider-saved.webp)


Done right you will see: **QCode · Open AI Compatible** ① is listed and turned on, and clicking **← Back** ② brings you back to **New Conversation**.




### Pick QCode in the model selector

In **New Conversation**, Msty Studio usually selects your first saved model automatically. Click the model selector at the bottom ① to open the picker, and click `gpt-6.2-sol` ② under **QCode** to confirm it is active.

![Back in New Conversation, open the bottom model selector and choose gpt-6.2-sol under QCode](https://docs.qcode.cc/static/images/tut/msty/202611/en/09-model-selector.webp)

**Stuck? The selector still says “Select a model” or QCode is missing**

Open **Settings → Model Providers**, make sure the toggle on the **QCode** card is switched on, then open the card and verify that at least one model checkbox (such as `gpt-6.2-sol`) is checked before clicking **Save**.



Done right you will see: the bottom selector ① displays `gpt-6.2-sol`, and the popover lists it under the **QCode** ② heading.




### Send a test message

Type `Hello, introduce yourself in one sentence.` into the chat box and press **Enter** to send ①. Wait a few seconds for `gpt-6.2-sol` to stream a reply ②.

![Send a test message and receive a normal reply from gpt-6.2-sol via QCode, confirming the setup works](https://docs.qcode.cc/static/images/tut/msty/202611/en/10-chat-reply.webp)


Done right you will see: your prompt ① appears on the right and a streamed reply ② from `gpt-6.2-sol` appears below it with a token speed indicator (`tok/s`) and no error banner.




## Verify it works

Your setup is complete when all three of these are true:

- **Settings → Model Providers** lists **QCode · Open AI Compatible** with its toggle enabled and at least `1 Model` selected.
- The model selector at the bottom of **New Conversation** shows `gpt-6.2-sol` under **QCode**.
- Sending a test message streams a normal reply with token speed (`tok/s`) and no error banner.


### All done

You can now chat with QCode models inside Msty Studio on Windows 11. From here you can:

- [Endpoints and API paths](/docs/getting-started/endpoints-and-api-paths) — OpenAI and Anthropic routes plus every QCode host
- [Model selection](/docs/usage/model-selection) — current GPT, Claude, and other model IDs
- [Billing](/docs/reference/billing) — how token usage is calculated
- [Troubleshooting](/docs/reference/troubleshooting) — fixing connection and authentication issues
- Compare plans on the [pricing page](https://qcode.cc/en/pricing?utm_source=docs.qcode.cc&utm_medium=tut_done&utm_campaign=knowledge_base&src=docs-ide-msty-done)
- Return to your [Dashboard](https://qcode.cc/en/dashboard) to check usage and balance


## Common errors

| Error or symptom | Why | What to do |
|---|---|---|
| `model_not_available_on_endpoint` | A Claude model was added under **Open AI Compatible** (`https://api.qcode.cc/openai/v1`), which does not serve Claude | Use GPT or Chinese models on **Open AI Compatible**, or add a separate built-in **Claude** provider using `https://api.qcode.cc/api/v1` |
| `404` | **Inference Endpoint** includes `/chat/completions`, or the built-in **Claude** provider was set to `https://api.qcode.cc/api` without `/v1` | Set **Inference Endpoint** to exactly `https://api.qcode.cc/openai/v1` for **Open AI Compatible**, or `https://api.qcode.cc/api/v1` for the built-in **Claude** provider, then click **Save** |
| `Model Hub > Model Providers` | Following an older Msty 2.x tutorial; Msty Studio 3.x reorganized the navigation menu | Confirm you installed Msty Studio 3.x, click **Settings** at the bottom of the left sidebar, and open **Model Providers** under **PROVIDERS & ENGINES** |
| `Some provider types and options require Aurum` | Picking an Aurum-only provider type (such as Azure or Amazon Bedrock) instead of **Open AI Compatible** | Open the **Provider** dropdown, search for `compatible`, and choose **Open AI Compatible** (or the built-in **Claude** provider) — neither requires Aurum |
| `Select a model` | No model checkbox was checked inside the **QCode** provider drawer, or the provider toggle in **Model Providers** is switched off | Open **Settings → Model Providers**, make sure **QCode** is toggled on, edit the provider, check `gpt-6.2-sol` (or add it via **+ Add Custom Model**), and click **Save** |

## FAQ

### What happens when my QCode API key expires?

Renewal depends on how you purchased the key. A key from a **personal plan** cannot be renewed after it expires — purchase a new plan, create a new key in your [Dashboard](https://qcode.cc/en/dashboard), and replace the **API Key** inside your **QCode** provider in Msty Studio. A key from a **pay-as-you-go pack** resumes working automatically as soon as you buy another pack, so you do not need to change anything in Msty Studio.

### Do I need a Msty account or an Aurum license to use QCode?

No. Msty Studio 3.0.1 launches with **No account needed.** and a **Skip to app →** link, and both **Open AI Compatible** and the built-in **Claude** provider work without an Aurum license.

### Can I use Claude models inside the Open AI Compatible provider?

No. QCode serves Claude models only on the Anthropic Messages route, so entering a Claude model ID under `https://api.qcode.cc/openai/v1` returns `model_not_available_on_endpoint`. Add a second provider using Msty Studio's built-in **Claude** type as described below.

### How do I change Msty Studio's interface language?

Open **Settings → General → Language** to choose from Msty Studio's downloadable locale packs or import a custom locale file.

## Optional: use Claude models with the built-in Claude provider

Msty Studio 3.0.1's built-in **Claude** provider includes an editable **Inference Endpoint** field (defaulting to `https://api.anthropic.com/v1/`), so you can run Claude models alongside your **Open AI Compatible** provider using the same QCode `cr_` key:

1. Open **Settings → Model Providers** and click **+ Add Provider** (in the top-right corner once at least one provider exists).
2. In the **Provider** dropdown, choose **Claude**.
3. Paste your QCode `cr_` key into **API Key**.
4. Change **Inference Endpoint (must start with http:// or https://)** to `https://api.qcode.cc/api/v1` (keep `/v1` to match Msty's built-in Claude path).
5. Click **Fetch Models**, check `claude-opus-5-5` and/or `claude-opus-5-5`, name the provider `QCode Claude`, and click **Save**.

| Field | Fill in |
|---|---|
| Provider | `Claude` |
| API Key | `cr_xxxxxxxx` |
| Inference Endpoint | `https://api.qcode.cc/api/v1` |
| Recommended models | `claude-opus-5-5` |

## Regional endpoints and other models

If you prefer a regional route, replace `api.qcode.cc` with `asia.qcode.cc` or `us.qcode.cc` and keep the exact path unchanged (`https://asia.qcode.cc/openai/v1` or `https://us.qcode.cc/openai/v1` for **Open AI Compatible**; `https://asia.qcode.cc/api/v1` or `https://us.qcode.cc/api/v1` for the built-in **Claude** provider). All three hosts offer the same models and differ only in network routing.

Besides `gpt-6.2-sol`, you can check or add other OpenAI-compatible models such as `gpt-6.2-sol`, `glm-5.3`, `kimi-k3`, and `deepseek-v4-pro`. If a model is not shown after clicking **Fetch Models**, click **+ Add Custom Model** above the model search box and enter the exact ID from [Model selection](/docs/usage/model-selection).

## Related docs

- [Endpoints and API paths](/docs/getting-started/endpoints-and-api-paths) — OpenAI and Anthropic routes across `api.qcode.cc`, `asia.qcode.cc`, and `us.qcode.cc`
- [Model selection](/docs/usage/model-selection) — current GPT, Claude, and Chinese model IDs
- [Billing](/docs/reference/billing) — token billing and usage tracking
- [Payment methods](/docs/reference/payment-methods) — supported payment options
- [Troubleshooting](/docs/reference/troubleshooting) — resolving common API and key issues