Skip to main content

​
Overview

Fireworks AI provides fast, reliable inference for open-source and proprietary language models with a developer-first approach and competitive pricing. Key Features:
  • Sub-second latency with optimized inference stack
  • 100+ models including Llama, Mixtral, and proprietary options
  • Fully OpenAI-compatible API
  • Fine-tuning and model deployment capabilities
Official Documentation: Fireworks AI Docs

​
Authentication

Fireworks AI uses Bearer token authentication with the OpenAI-compatible format. Header: 
Authorization: Bearer YOUR_FIREWORKS_API_KEY
Lava Forward Token: 
${LAVA_SECRET_KEY}.${CONNECTION_SECRET}.${PRODUCT_SECRET}
For BYOK (Bring Your Own Key): 
${LAVA_SECRET_KEY}.${CONNECTION_SECRET}.${PRODUCT_SECRET}.${YOUR_FIREWORKS_API_KEY}
ModelContextDescriptionUse Case
accounts/fireworks/models/llama-v3p3-70b-instruct128KMeta’s Llama 3.3 flagshipGeneral reasoning, coding
accounts/fireworks/models/mixtral-8x7b-instruct32KMistral’s MoE modelFast, balanced performance
accounts/fireworks/models/qwen2p5-72b-instruct128KAlibaba’s Qwen 2.5Multilingual, math
Pricing: See Fireworks AI Pricing for current rates.

​
Quick Start Example

// 1. Set up your environment variables
const LAVA_FORWARD_TOKEN = process.env.LAVA_FORWARD_TOKEN;

// 2. Define the Fireworks AI endpoint
const FIREWORKS_ENDPOINT = 'https://api.fireworks.ai/inference/v1/chat/completions';

// 3. Make the request through Lava
const response = await fetch(
  `https://api.lavapayments.com/v1/forward?u=${encodeURIComponent(FIREWORKS_ENDPOINT)}`,
  {
    method: 'POST',
    headers: {
      'Authorization': `Bearer ${LAVA_FORWARD_TOKEN}`,
      'Content-Type': 'application/json'
    },
    body: JSON.stringify({
      model: 'accounts/fireworks/models/llama-v3p3-70b-instruct',
      messages: [
        {
          role: 'user',
          content: 'What makes a good AI inference platform?'
        }
      ],
      temperature: 0.7,
      max_tokens: 512
    })
  }
);

// 4. Parse response and extract usage
const data = await response.json();
console.log('Response:', data.choices[0].message.content);

// 5. Track usage (from response body)
const usage = data.usage;
console.log('Tokens used:', usage.total_tokens);

// 6. Get Lava request ID (from headers)
const requestId = response.headers.get('x-lava-request-id');
console.log('Lava Request ID:', requestId);

​
Available Endpoints

Fireworks AI supports OpenAI-compatible endpoints:
EndpointMethodDescription
/inference/v1/chat/completionsPOSTText generation with conversation context
/inference/v1/modelsGETList available models

​
Usage Tracking

Usage data is returned in the response body (OpenAI format): 
{
  "usage": {
    "prompt_tokens": 30,
    "completion_tokens": 200,
    "total_tokens": 230
  }
}
Location: data.usage Format: Standard OpenAI usage object Lava Tracking: Automatically tracked via x-lava-request-id header

​
Features & Capabilities

Streaming: 
{
  "stream": true
}
JSON Mode: 
{
  "response_format": { "type": "json_object" }
}
Function Calling: 
{
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "search",
        "description": "Search the web",
        "parameters": {
          "type": "object",
          "properties": {
            "query": { "type": "string" }
          }
        }
      }
    }
  ]
}

​
BYOK Support

Status: ✅ Supported (managed keys + BYOK) BYOK Implementation:
  • Append your Fireworks AI API key to the forward token: ${TOKEN}.${YOUR_FIREWORKS_KEY}
  • Lava tracks usage and billing while you maintain key control
  • No additional Lava API key costs (metering-only mode available)
Getting a Fireworks AI API Key:
  1. Sign up at Fireworks AI Console
  2. Navigate to API Keys section
  3. Create a new API key
  4. Use in Lava forward token (4th segment)

​
Best Practices

  1. Model Selection: Use Llama 3.3 for reasoning, Mixtral for speed/cost, Qwen for multilingual
  2. Model Names: Use full account paths (e.g., accounts/fireworks/models/llama-v3p3-70b-instruct)
  3. Temperature: 0.7 for creative tasks, 0.1-0.3 for factual outputs
  4. Context Management: Leverage 128K context models for long-form content
  5. Error Handling: Fireworks returns standard OpenAI error formats

​
Performance Characteristics

Latency: Sub-second first-token latency for most models Throughput: Optimized for high-concurrency workloads Reliability: 99.9% uptime SLA for production deployments Use Cases:
  • Production chatbots
  • Content generation pipelines
  • Code assistance tools
  • Multi-modal applications

​
Additional Resources