> 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/gpu-pricing.md).

# GPU Pricing

Detailed pricing information for all available GPU types and use cases.

## GPU Specifications

### High-End GPUs

#### H200 141GB

* **Price**: $0.047/minute
* **Memory**: 141GB HBM3e
* **Use cases**: Largest models, maximum performance
* **Best for**: Enterprise applications, research

#### H100 SXM 80GB

* **Price**: $0.044/minute
* **Memory**: 80GB HBM3
* **Use cases**: Large models, high-performance inference
* **Best for**: Production workloads, large-scale deployments

### Mid-Range GPUs

#### A100 PCIe 80GB

* **Price**: $0.024/minute
* **Memory**: 80GB HBM2e
* **Use cases**: Large models, good performance/cost ratio
* **Best for**: Most production use cases

#### L40S 48GB

* **Price**: $0.013/minute
* **Memory**: 48GB GDDR6
* **Use cases**: Medium models, good performance
* **Best for**: Balanced workloads

### Cost-Effective GPUs

#### A6000 48GB

* **Price**: $0.009/minute
* **Memory**: 48GB GDDR6
* **Use cases**: Medium models, cost-effective
* **Best for**: Development, testing

#### 4090 24GB

* **Price**: $0.007/minute
* **Memory**: 24GB GDDR6X
* **Use cases**: Smaller models, most cost-effective
* **Best for**: Development, small-scale production

## Cost Comparison

### Hourly Rates

| GPU            | Per Hour | Per Day | Per Month\* |
| -------------- | -------- | ------- | ----------- |
| H200 141GB     | $2.82    | $67.68  | $2,030      |
| H100 SXM 80GB  | $2.64    | $63.36  | $1,901      |
| A100 PCIe 80GB | $1.44    | $34.56  | $1,037      |
| L40S 48GB      | $0.78    | $18.72  | $562        |
| A6000 48GB     | $0.54    | $12.96  | $389        |
| 4090 24GB      | $0.42    | $10.08  | $302        |

\*Based on 24/7 usage

## Choosing the Right GPU

### Model Size Guidelines

#### Small Models (<7B parameters)

* **Recommended**: 4090 24GB
* **Cost**: $0.42/hour
* **Performance**: Excellent for small models

#### Medium Models (7B-13B parameters)

* **Recommended**: A6000 48GB or L40S 48GB
* **Cost**: $0.54-0.78/hour
* **Performance**: Good balance of cost and performance

#### Large Models (13B-70B parameters)

* **Recommended**: A100 PCIe 80GB
* **Cost**: $1.44/hour
* **Performance**: Excellent for large models

#### Very Large Models (70B+ parameters)

* **Recommended**: H100 SXM 80GB or H200 141GB
* **Cost**: $2.64-2.82/hour
* **Performance**: Maximum performance for largest models

### Use Case Recommendations

#### Development & Testing

* **Start with**: 4090 24GB
* **Upgrade to**: A6000 48GB if needed
* **Cost optimization**: Use only when needed

#### Production Workloads

* **Small scale**: A6000 48GB
* **Medium scale**: A100 PCIe 80GB
* **Large scale**: H100 SXM 80GB
* **Enterprise**: H200 141GB

#### Research & Experimentation

* **Budget conscious**: 4090 24GB
* **Performance focused**: A100 PCIe 80GB
* **Maximum performance**: H200 141GB

## Cost Optimization Tips

### 1. Right-Size Your GPU

* **Start small**: Begin with 4090 or A6000
* **Monitor performance**: Track actual usage
* **Scale appropriately**: Upgrade only when needed

### 2. Usage Patterns

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

### 3. Tier Benefits

* **Tier 2**: Access to 4090 and A6000
* **Tier 3**: Access to L40S and A100
* **Tier 4**: Access to all GPUs including H200/H100

## Billing and Usage

### Billing Model

* **Per-minute billing**: Pay only for active usage
* **Automatic scaling**: Start/stop endpoints as needed
* **Usage tracking**: Monitor costs in real-time
* **Cost alerts**: Set spending limits and notifications

### Usage Monitoring

* **Real-time metrics**: Track GPU utilization
* **Cost tracking**: Monitor spending by endpoint
* **Performance metrics**: Track model performance
* **Usage reports**: Detailed billing information

## Getting Started

1. **Choose your tier**: Ensure you have access to desired GPUs
2. **Select GPU type**: Based on your model size and performance needs
3. **Deploy model**: Upload and configure your model
4. **Monitor usage**: Track costs and performance
5. **Optimize**: Adjust GPU type based on actual needs