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

# Performance and Limits

> Rate limits, service tiers, and infrastructure information

# Performance and Limits

Understand rate limits, service tiers, and infrastructure capabilities for Hyperbolic's Serverless Inference API.

## Rate Limits

### Standard Limits

| Tier           | Requests/Minute | Requirements  |
| -------------- | --------------- | ------------- |
| **Basic**      | 60              | Free account  |
| **Pro**        | 600             | \$5+ deposit  |
| **Enterprise** | Unlimited       | Contact sales |

<Info>
  All tiers have a per-IP limit of 600 requests/minute for DDoS protection.
</Info>

###

### Upgrading to Pro

Get 10x higher rate limits by upgrading to Pro:

<Steps>
  <Step title="Log into Dashboard">
    Go to [app.hyperbolic.ai](https://app.hyperbolic.ai) and sign in.
  </Step>

  <Step title="Add Funds">
    Deposit \$5 or more to your account.
  </Step>

  <Step title="Automatic Upgrade">
    Your account is automatically upgraded to Pro tier.
  </Step>
</Steps>

## Service Tiers

| Feature                 | Basic             | Pro          | Enterprise     |
| ----------------------- | ----------------- | ------------ | -------------- |
| **Rate Limit**          | 60/min            | 600/min      | Unlimited      |
| **Cost**                | Free              | \$5+ deposit | Custom         |
| **Support**             | Community Discord | Email        | 24/7 dedicated |
| **Priority Queue**      | -                 | Yes          | Yes            |
| **Dedicated Instances** | -                 | -            | Yes            |
| **Custom SLAs**         | -                 | -            | Yes            |
| **Fine-tuning**         | -                 | -            | Yes            |

<Tip>
  Need higher limits or dedicated infrastructure? [Contact sales](mailto:sales@hyperbolic.ai)
</Tip>

## Pricing Summary

Hyperbolic uses pay-as-you-go pricing with no monthly quotas or commitments.

### Text Generation

| Model Category           | Price                         |
| ------------------------ | ----------------------------- |
| Small models (3B-8B)     | From \$0.10 per 1M tokens     |
| Medium models (32B-72B)  | \$0.20 - \$0.40 per 1M tokens |
| Large models (120B-480B) | \$0.30 - \$4.00 per 1M tokens |

### Image Generation

**Base rate:** \$0.01 per image (1024x1024, 25 steps)

**Formula:** `$0.01 × (width/1024) × (height/1024) × (steps/25)`

### Audio Generation

**Rate:** \$5.00 per 1M characters

<Info>
  See [Text APIs](/docs/inference/text-apis), [Image APIs](/docs/inference/image-apis), and [Audio APIs](/docs/inference/audio-apis) for complete pricing by model.
</Info>

## Infrastructure

### Security

| Feature                 | Description                                           |
| ----------------------- | ----------------------------------------------------- |
| **Zero Data Retention** | Your prompts and responses are never stored           |
| **Encryption**          | TLS 1.3 for all API connections                       |
| **Compliance**          | We are currently pursuing SOC 2 Type II certification |

## Error Handling

### Rate Limit Errors

When you exceed rate limits, you'll receive a `429 Too Many Requests` response:

```json theme={null}
{
  "error": {
    "code": "rate_limit_exceeded",
    "message": "Rate limit exceeded. Please retry after X seconds."
  }
}
```

### Best Practices

* **Implement exponential backoff** for automatic retries
* **Monitor usage** via the dashboard to stay within limits
* **Cache responses** when appropriate to reduce API calls
* **Use streaming** for long responses to improve perceived latency

### Retry Example

```python theme={null}
import time
from openai import RateLimitError

def call_with_retry(func, max_retries=3):
    """Call a function with exponential backoff on rate limit errors."""
    for attempt in range(max_retries):
        try:
            return func()
        except RateLimitError:
            if attempt == max_retries - 1:
                raise
            wait_time = 2 ** attempt  # 1, 2, 4 seconds
            time.sleep(wait_time)
    raise Exception("Max retries exceeded")

# Usage
response = call_with_retry(
    lambda: client.chat.completions.create(
        model="meta-llama/Llama-3.3-70B-Instruct",
        messages=[{"role": "user", "content": "Hello!"}]
    )
)
```

## Monitoring Usage

Track your API usage in the [Hyperbolic Dashboard](https://app.hyperbolic.ai/usage):

* Requests per minute/hour/day
* Token consumption by model
* Cost breakdown and billing history
* Real-time usage graphs

## Next Steps

<CardGroup cols={3}>
  <Card title="Text APIs" icon="message" href="/docs/inference/text-apis">
    Models and pricing details
  </Card>

  <Card title="Image APIs" icon="image" href="/docs/inference/image-apis">
    Image generation pricing
  </Card>

  <Card title="Audio APIs" icon="volume-high" href="/docs/inference/audio-apis">
    Text-to-speech pricing
  </Card>
</CardGroup>