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

# Fireworks AI

> Configure Fireworks AI high-speed inference in EnConvo

## Overview

Fireworks AI is a high-performance inference platform optimized for speed. It serves open-source models with extremely low latency and high throughput, making it an excellent choice for real-time applications and production workloads.

## Supported Models

| Model               | Description            | Best For           |
| ------------------- | ---------------------- | ------------------ |
| **Llama 3.1 405B**  | Largest open model     | Complex reasoning  |
| **Llama 3.1 70B**   | Strong general model   | Analysis, coding   |
| **Llama 3.1 8B**    | Fast lightweight model | Quick responses    |
| **Mixtral 8x22B**   | Mistral MoE            | Balanced workloads |
| **Qwen 2.5 72B**    | Alibaba's model        | Multilingual tasks |
| **FireFunction V2** | Function calling model | Tool use, agents   |

## Setup

<Steps>
  <Step title="Get API Key">
    1. Go to [Fireworks AI](https://fireworks.ai)
    2. Sign in or create an account
    3. Navigate to **Account** → **API Keys**
    4. Create a new API key
  </Step>

  <Step title="Configure in EnConvo">
    1. Open **Settings** → **AI Provider**
    2. Select **Fireworks AI**
    3. Go to **Credentials** module
    4. Enter your API key
  </Step>

  <Step title="Select Model">
    Choose your preferred model from the dropdown
  </Step>
</Steps>

## Configuration

| Setting         | Description           | Default       |
| --------------- | --------------------- | ------------- |
| **Credentials** | API key configuration | Required      |
| **Model Name**  | Model to use          | Llama 3.1 70B |
| **Temperature** | Creativity (0-2)      | Medium (1)    |

## Validate and Use

<Steps>
  <Step title="Validate credentials">
    Click **Validate** in the Fireworks credential settings. If validation fails, confirm the API key is active and the account has available credits or quota.
  </Step>

  <Step title="Use exact model paths">
    Prefer selecting from the model dropdown. If you type a model manually, use the exact Fireworks model path from the Fireworks model catalog.
  </Step>

  <Step title="Test latency">
    Fireworks is optimized for fast inference. If a model is slow, test a smaller model or a different hosted variant.
  </Step>
</Steps>

## Pricing

Check [Fireworks AI Pricing](https://fireworks.ai/pricing) for current rates.

| Model          | Input     | Output    |
| -------------- | --------- | --------- |
| Llama 3.1 405B | \$3.00/1M | \$3.00/1M |
| Llama 3.1 70B  | \$0.90/1M | \$0.90/1M |
| Llama 3.1 8B   | \$0.20/1M | \$0.20/1M |
| Mixtral 8x22B  | \$1.20/1M | \$1.20/1M |

<Note>
  Fireworks AI offers a free tier with rate limits, so you can test models before adding payment.
</Note>

## Why Fireworks?

<CardGroup cols={2}>
  <Card title="Ultra-Fast Inference" icon="bolt">
    Optimized for lowest possible latency
  </Card>

  <Card title="Function Calling" icon="wrench">
    FireFunction models excel at tool use
  </Card>

  <Card title="OpenAI Compatible" icon="plug">
    Drop-in replacement for OpenAI API format
  </Card>

  <Card title="Scalable" icon="chart-line">
    Handles high-throughput production workloads
  </Card>
</CardGroup>

## Best Practices

<AccordionGroup>
  <Accordion title="Model Selection">
    * **Llama 3.1 70B**: Best default for most tasks -- fast and capable
    * **Llama 3.1 405B**: When you need the highest quality open-source output
    * **Llama 3.1 8B**: Real-time chat, high-volume processing
    * **FireFunction V2**: Agent workflows that require reliable function calling
  </Accordion>

  <Accordion title="Performance Tips">
    * Fireworks excels at low-latency responses -- ideal for interactive use cases
    * Use smaller models for streaming chat to minimize time-to-first-token
    * Monitor your usage in the Fireworks dashboard
  </Accordion>
</AccordionGroup>

## Troubleshooting

<AccordionGroup>
  <Accordion title="Invalid API key">
    * Verify the key is copied correctly from [fireworks.ai/account/api-keys](https://fireworks.ai/account/api-keys)
    * Check if your account has sufficient credits
    * Ensure the key has not been revoked
  </Accordion>

  <Accordion title="Rate limits">
    * Free tier has rate limits per minute
    * Upgrade to a paid plan for higher throughput
    * Use smaller models to reduce token consumption
  </Accordion>

  <Accordion title="Model not found">
    * Model identifiers may change -- check the Fireworks model catalog
    * Ensure you are using the full model path (e.g., `accounts/fireworks/models/llama-v3p1-70b-instruct`)
    * Try selecting the model from the EnConvo dropdown instead of typing manually
  </Accordion>
</AccordionGroup>