Together AI Provider

Access a broad catalog of open-source embedding and chat models through Together AI's inference platform.

Overview

Together AI hosts hundreds of open-source models with serverless and dedicated deployment options. This integration configures CortexDB to use Together AI for:

• Embedding generation — high-quality open-source embedding models
• Entity extraction — Llama, Mistral, and other chat models for relationship extraction

Installation

pip install cortexdbai[together]

Configuration

| Environment Variable | Default | Description | |---|---|---| | CORTEX_TOGETHER_API_KEY | Required | Together AI API key | | CORTEX_TOGETHER_EMBED_MODEL | togethercomputer/m2-bert-80M-8k-retrieval | Embedding model | | CORTEX_TOGETHER_CHAT_MODEL | meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo | Chat model for entity extraction |

Embedding Models

| Model | Dimensions | Description | |---|---|---| | togethercomputer/m2-bert-80M-8k-retrieval | 768 | Fast, lightweight retrieval model | | BAAI/bge-large-en-v1.5 | 1024 | High-quality general-purpose embeddings | | WhereIsAI/UAE-Large-V1 | 1024 | Strong performance on MTEB benchmarks |

Usage Example

from cortexdb_together import TogetherEmbeddingProvider, TogetherConfig

config = TogetherConfig(api_key="your-together-api-key")

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

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

Using a Different Model

config = TogetherConfig(
    api_key="your-together-api-key",
    embed_model="BAAI/bge-large-en-v1.5",
)
provider = TogetherEmbeddingProvider(config=config)
# provider.dimension == 1024

Switching Providers

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

from cortexdb_together import TogetherEmbeddingProvider

provider = TogetherEmbeddingProvider()  # 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 Together AI provider, the SDK translates your calls into REST API requests against the CortexDB and Together AI endpoints.

Storing a memory (remember)

# SDK: cortex.remember("CQRS separates reads from writes.")
curl -X POST https://api.cortexdb.ai/v1/remember \
  -H "Authorization: Bearer $CORTEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "CQRS separates reads from writes.",
    "tenant_id": "my-app"
  }'
# Returns: { "event_id": "evt_abc123" }

Retrieving context (recall)

# SDK: result = cortex.recall("What is CQRS?")
# 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 is CQRS?",
    "tenant_id": "my-app"
  }'
# Returns: { "context": "...", "confidence": 0.91, "latency_ms": 52 }

Generating embeddings (Together AI)

# The provider calls Together AI's embedding endpoint
curl -X POST https://api.together.xyz/v1/embeddings \
  -H "Authorization: Bearer $TOGETHER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "togethercomputer/m2-bert-80M-8k-retrieval",
    "input": "What is CQRS?"
  }'
# Returns: { "data": [{ "embedding": [0.123, -0.456, ...] }] }