> For the complete documentation index, see [llms.txt](https://docs.dagploy.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dagploy.com/home/blog/blog-post/build-claude-alternative-in-20-mins.md).

# Build Claude Alternative in 20 Mins

Running DAX, OpenWork and GPT OSS 20B in GCP

Here are few steps to setup and run Claude alternative using OSS that runs in your cloud. Everything full in your control without any external API or sharing data required.

{% embed url="<https://www.youtube.com/watch?v=vI6Oqr1OpI8>" %}

### 1. Install DAX

This provide automated GPU instance provisioning in our cloud without dealing with complexity like NVIDIA drivers, stuck in downloading large models, docker setup and many others things. This step will required 5 minutes of setup and make sure you have `GPU Quota` activated in your GCP project.

Go to <https://github.com/dagploy/dax> and run the installation

### 2. Download GPT OSS 20B and VLLM

Start by caching Docker images and models first, around 100GB in total.

#### Step 1: Cache the VLLM docker <a href="#step-1-cache-the-vllm-docker" id="step-1-cache-the-vllm-docker"></a>

<a class="button secondary">Ask</a><a class="button secondary">Copy</a>

```
dax run download_docker vllm/vllm-openai:nightly,ghcr.io/open-webui/open-webui:main --images vllm-lib --image-size 100
```

#### Step 2: Cache GPTOSS 20B from Huggingface <a href="#step-2-cache-gptoss-20b-from-huggingface" id="step-2-cache-gptoss-20b-from-huggingface"></a>

<a class="button secondary">Ask</a><a class="button secondary">Copy</a>

```
dax run download_hf openai/gpt-oss-20b --image-size 50
```

You can start tunneling from host into your GCP VM or using VPN.

### Step 3: Run inferencing <a href="#run-inferencing" id="run-inferencing"></a>

<a class="button secondary">Ask</a><a class="button secondary">Copy</a>

```
dax run create_vm_inference --stack-name gptoss --config-json '{"images":["models--openai
```

### 3. Install OpenWork in your Laptop

Download Openworks&#x20;

{% embed url="<https://github.com/different-ai/openwork/releases>" %}

Connect openwork with your GPT OSS 20B VM instance. Follow their tutorial. For Linux, you need to edit `vim ~/.config/opencode/opencode.json`&#x20;

Make sure to match the `model` , `port` and `url`&#x20;

```
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "my-api": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "GPT OSS 20B",
      "options": {
        "baseURL": "http://localhost:8000/v1"
      },
      "models": {
        "model-name": {
          "name": "model"
        }
      }
    }
  }
}

```

Author tested that **Qwen 3.6 27B** [**https://huggingface.co/Qwen/Qwen3.6-27B**](https://huggingface.co/Qwen/Qwen3.6-27B)  perform better on integration with OpenWork.

Do you need assistance to build Sovereign AI infra in your organization? feel free to drop message at [**https://www.dagploy.com/contact**](https://www.dagploy.com/contact)

Or, learn more about Dagploy&#x20;

{% embed url="<https://www.dagploy.com>" %}

&#x20;