> For the complete documentation index, see [llms.txt](https://docs.meganova.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.meganova.ai/compute/dedicated-endpoints.md).

# Dedicated Endpoints

When hosting your own model you pay per minute for the GPU endpoints, whether it is a model you fine-tuned or any other model you choose to host. You can start or stop your endpoint any time through the web-based portal.

## Your Fine-Tuned Models

### GPU Pricing (Per Minute)

| GPU TYPE         | PRICE PER MINUTE HOSTED |
| ---------------- | ----------------------- |
| 1×H200 141GB     | $0.047                  |
| 1×H100 SXM 80GB  | $0.044                  |
| 1×A100 PCIe 80GB | $0.024                  |
| 1×L40S 48GB      | $0.013                  |
| 1×A6000 48GB     | $0.009                  |
| 1×4090 24GB      | $0.007                  |

## Getting Started

### 1. Choose Your GPU

* **H200 141GB**: Highest performance, best for large models
* **H100 SXM 80GB**: Excellent performance, good for most use cases
* **A100 PCIe 80GB**: Great balance of performance and cost
* **L40S 48GB**: Good for medium-sized models
* **A6000 48GB**: Cost-effective for smaller models
* **4090 24GB**: Most affordable option

### 2. Deploy Your Model

1. **Upload your model** to the platform
2. **Select GPU type** based on your needs
3. **Configure endpoint** settings
4. **Start deployment** - typically takes 2-5 minutes

### 3. Manage Your Endpoint

* **Start/Stop**: Control your endpoint anytime
* **Monitor usage**: Track costs and performance
* **Scale up/down**: Change GPU type as needed
* **Access logs**: Monitor model performance

## Use Cases

### Fine-Tuned Models

* **Custom training** on your data
* **Domain-specific** applications
* **Specialized tasks** not covered by general models

### Custom Models

* **Open-source models** not available in serverless
* **Experimental models** for testing
* **High-performance** inference needs

### Production Workloads

* **Consistent latency** requirements
* **High-volume** inference
* **Custom model** serving

## Cost Optimization

### Right-Sizing Your GPU

* **Start small**: Begin with 4090 or A6000
* **Monitor usage**: Track actual performance needs
* **Scale up**: Upgrade GPU if needed
* **Scale down**: Downgrade to save costs

### Usage Patterns

* **Development**: Use cheaper GPUs for testing
* **Production**: Use appropriate GPU for performance
* **Batch processing**: Start/stop as needed
* **Scheduled workloads**: Use automation