> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gamerhash.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Install the GamerHash AI App

> Step-by-step: from download to first chat, in five minutes.

export const Win = () => <span style={{
  display: 'inline-block',
  padding: '2px 10px',
  borderRadius: '999px',
  fontSize: '0.82em',
  fontWeight: 600,
  background: 'rgba(0,120,212,0.18)',
  color: '#7CC4FF',
  border: '1px solid #0078D4',
  whiteSpace: 'nowrap'
}}>Windows</span>;

This walks you from downloading the AI App to your first chat completion. Time: **5 minutes** plus model download (which happens in the background and varies by model size).

## Before you start

<CardGroup cols={2}>
  <Card title="Hardware check" icon="microchip">
    <Win /> 10/11, NVIDIA RTX with **8 GB+ VRAM**, recent driver. Full breakdown in [Hardware requirements](/ai/hardware-requirements).
  </Card>

  <Card title="Disk space" icon="hard-drive">
    \~30 GB free for downloaded models. More if you keep many models cached.
  </Card>

  <Card title="GamerHash account" icon="user">
    Created during install — email or Google login.
  </Card>

  <Card title="Network" icon="wifi">
    Broadband. Initial model downloads are large (4–20 GB depending on model).
  </Card>
</CardGroup>

## 1. Download

<Steps>
  <Step title="Open the download page">
    Go to [gamerhash.com/en/app-download](https://gamerhash.com/en/app-download) and click **Download GamerHash AI**. The installer is `.exe` and signed.
  </Step>

  <Step title="Run the installer">
    Double-click the downloaded `.exe`. Windows may ask for admin permission — accept. The installer is signed by GamerHash; if Windows SmartScreen warns, click **More info → Run anyway**.
  </Step>

  <Step title="Pick install path">
    Default is fine for most users. If you have multiple drives, pick the one with the most free space — model files end up here.
  </Step>
</Steps>

## 2. First run

<Steps>
  <Step title="Sign in or create an account">
    On first launch you'll be prompted. **Sign in with Google** is the fastest path. Email + password also works.
  </Step>

  <Step title="Run the GPU benchmark">
    The app benchmarks your GPU once to size workloads correctly. Takes \~1 minute. You won't need to repeat this unless you change hardware.
  </Step>

  <Step title="Pick your first model">
    The Model Center opens. For a first chat, pick any model marked as a fit for your GPU (each model shows a VRAM-fit badge). Click **Download** — progress shows in the bar.
  </Step>
</Steps>

<Note>
  First model download is the slowest step. Smaller text models (\~5 GB) download in 5–15 minutes on a typical home connection. Image and video models can be 10–20 GB.
</Note>

## 3. Your first chat

<Steps>
  <Step title="Open the Chat module">
    From the dashboard, click **Chat**. The model you just downloaded should be selected.
  </Step>

  <Step title="Send a prompt">
    Type a question and hit Enter. First response includes a model warmup — subsequent responses are faster.
  </Step>

  <Step title="Inspect the response">
    The response streams token-by-token. Generation runs entirely on your GPU; nothing is sent to a third-party cloud.
  </Step>
</Steps>

## 4. Try image generation

<Steps>
  <Step title="Switch to the Image module">
    From the dashboard sidebar, pick **Image Generation**.
  </Step>

  <Step title="Pick a model">
    For 8 GB GPUs, **Sana Sprint** is a good first choice — fast and high quality. For 12 GB+, try **FLUX.1 Schnell**.
  </Step>

  <Step title="Generate">
    Type a prompt, click **Generate**. Output appears in the gallery; right-click to save or copy. Images are stored locally only.
  </Step>
</Steps>

## 5. Opt in to earn (optional)

If you want to share idle GPU power for GUSD:

<Steps>
  <Step title="Open the earning module">
    Sidebar → **Earning**. Read the explanation, then toggle the module on.
  </Step>

  <Step title="Leave it on when you're not gaming">
    Earning workloads take the full GPU and VRAM, so you can't earn and game at the same time. Launching a game pushes the module out automatically; close the game and earning resumes on its own.
  </Step>

  <Step title="Watch GUSD accrue">
    Workloads start dispatching to your GPU as the queue allows. The dashboard shows live earnings per hour.
  </Step>
</Steps>

## Troubleshooting

<AccordionGroup>
  <Accordion title="Installer won't run / Windows SmartScreen warning">
    Click **More info** → **Run anyway**. The installer is signed by GamerHash sp. z o.o. — verify the publisher in the certificate dialog.
  </Accordion>

  <Accordion title="Benchmark fails">
    Almost always a driver issue. Update to the latest **NVIDIA Studio** driver, reboot, retry. If it still fails, try the **Game Ready** driver. Report persistent failures in Discord `#support` with your GPU model.
  </Accordion>

  <Accordion title="Model download is stuck">
    Pause and resume from the Model Center. If it keeps stalling, check disk space and disable any aggressive antivirus. Some AVs scan multi-GB files extremely slowly.
  </Accordion>

  <Accordion title="Chat is slow / first response delayed">
    First response loads the model into VRAM (cold start). Subsequent responses are fast. If chat is consistently slow, your GPU may be hitting VRAM limits — switch to a smaller quantized variant of the same model.
  </Accordion>

  <Accordion title="The earning module never gets workloads">
    Possible reasons: too few hours of uptime, GPU class limits the workload pool, network demand is low at the moment. Check the in-app dashboard and Medium community updates for current network demand.
  </Accordion>
</AccordionGroup>

## Next

<CardGroup cols={2}>
  <Card title="Earn with GPU sharing" icon="coins" href="/tutorials/earn-with-gpu">
    Deeper walk-through of the earning module.
  </Card>

  <Card title="Modules" icon="layer-group" href="/ai/modules">
    Tour every module the app ships.
  </Card>
</CardGroup>