Fireworks AI Provider

Run CortexDB's embedding and entity-extraction pipelines on Fireworks AI's optimized inference infrastructure.

Overview

Fireworks AI provides fast, cost-efficient inference for open-source models with production-grade reliability. This integration configures CortexDB to use Fireworks AI for:

• Embedding generation — optimized open-source embedding models
• Entity extraction — fast chat model inference for relationship extraction

Installation

pip install cortexdbai[fireworks]

Configuration

| Environment Variable | Default | Description | |---|---|---| | CORTEX_FIREWORKS_API_KEY | Required | Fireworks AI API key | | CORTEX_FIREWORKS_EMBED_MODEL | nomic-ai/nomic-embed-text-v1.5 | Embedding model | | CORTEX_FIREWORKS_CHAT_MODEL | accounts/fireworks/models/llama-v3p1-70b-instruct | Chat model |

Embedding Models

| Model | Dimensions | Description | |---|---|---| | nomic-ai/nomic-embed-text-v1.5 | 768 | High-quality, long-context embedding model by Nomic | | thenlper/gte-large | 1024 | General Text Embeddings by Alibaba DAMO |

Usage Example

from cortexdb_fireworks import FireworksEmbeddingProvider, FireworksConfig

config = FireworksConfig(api_key="your-fireworks-api-key")

async with FireworksEmbeddingProvider(config=config) as provider:
    embedding = await provider.embed_query("What is event sourcing?")
    print(f"Dimension: {provider.dimension}")  # 768

    embeddings = await provider.embed([
        "Event sourcing stores all changes as events.",
        "CQRS separates reads from writes.",
    ])

Using a Different Model

config = FireworksConfig(
    api_key="your-fireworks-api-key",
    embed_model="thenlper/gte-large",
)
provider = FireworksEmbeddingProvider(config=config)
# provider.dimension == 1024

Switching Providers

To switch CortexDB from the default OpenAI embeddings to Fireworks AI:

from cortexdb_fireworks import FireworksEmbeddingProvider

provider = FireworksEmbeddingProvider()  # reads from env vars

All CortexDB embedding providers implement the same interface (embed, embed_query, dimension, model_name), so switching is a one-line change.

Under the Hood

When using the Fireworks AI provider, the SDK translates your calls into REST API requests against the CortexDB and Fireworks AI endpoints.

Storing a memory (remember)

# SDK: cortex.remember("The payments service was migrated to CockroachDB.")
curl -X POST https://api.cortexdb.ai/v1/remember \
  -H "Authorization: Bearer $CORTEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "The payments service was migrated to CockroachDB.",
    "tenant_id": "my-app"
  }'
# Returns: { "event_id": "evt_abc123" }

Retrieving context (recall)

# SDK: result = cortex.recall("What database does the payments service use?")
# result.context, result.confidence, result.latency_ms
curl -X POST https://api.cortexdb.ai/v1/recall \
  -H "Authorization: Bearer $CORTEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What database does the payments service use?",
    "tenant_id": "my-app"
  }'
# Returns: { "context": "...", "confidence": 0.94, "latency_ms": 41 }

Generating embeddings (Fireworks AI)

# The provider calls Fireworks AI's embedding endpoint
curl -X POST https://api.fireworks.ai/inference/v1/embeddings \
  -H "Authorization: Bearer $FIREWORKS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nomic-ai/nomic-embed-text-v1.5",
    "input": "What database does the payments service use?"
  }'
# Returns: { "data": [{ "embedding": [0.123, -0.456, ...] }] }