Skip to content

Fireworks AI¶

Fireworks AI provides the fastest inference speeds for open-source models with enterprise reliability.

Quick Start¶

from sentimatrix import Sentimatrix
from sentimatrix.config import SentimatrixConfig, LLMConfig

config = SentimatrixConfig(
    llm=LLMConfig(
        provider="fireworks",
        model="accounts/fireworks/models/llama-v3p1-70b-instruct",
        api_key="your-fireworks-key"  # Or set FIREWORKS_API_KEY
    )
)

async with Sentimatrix(config) as sm:
    summary = await sm.summarize_reviews(reviews)

Available Models¶

Model	Context	Best For
`accounts/fireworks/models/llama-v3p1-405b-instruct`	128K	Highest quality
`accounts/fireworks/models/llama-v3p1-70b-instruct`	128K	Balanced
`accounts/fireworks/models/llama-v3p1-8b-instruct`	128K	Fast
`accounts/fireworks/models/mixtral-8x22b-instruct`	64K	MoE
`accounts/fireworks/models/qwen2-72b-instruct`	128K	Multilingual
`accounts/fireworks/models/firefunction-v2`	8K	Function calling

Configuration¶

LLMConfig(
    provider="fireworks",
    model="accounts/fireworks/models/llama-v3p1-70b-instruct",
    api_key="your-key",           # Or FIREWORKS_API_KEY env var
    temperature=0.7,
    max_tokens=4096,
    timeout=30,
)

Environment Variables¶

export FIREWORKS_API_KEY="your-fireworks-api-key"

Features¶

Fastest Inference: Industry-leading speed
Function Calling: FireFunction models
Batch API: High-throughput batch processing
OpenAI Compatible: Drop-in replacement

Pricing (Examples)¶

Model	Input	Output
Llama 3.1 405B	$3/1M	$3/1M
Llama 3.1 70B	$0.90/1M	$0.90/1M
Llama 3.1 8B	$0.20/1M	$0.20/1M
Mixtral 8x22B	$0.90/1M	$0.90/1M

Example: High-Throughput Processing¶

async with Sentimatrix(config) as sm:
    # Fireworks excels at batch processing
    reviews = [...]  # 1000s of reviews

    results = await sm.analyze_batch(
        reviews,
        batch_size=100  # Large batches supported
    )