Hosted Model Inference - Beam

Use hosted models from the dashboard’s Endpoints catalog.

Discover Models

curl --fail-with-body https://app.beam.cloud/v1/models \
  -H "Authorization: Bearer $BEAM_TOKEN"

The catalog is public. Include a token to see models available to your workspace.

Call a Chat Model

Install the OpenAI Python client and set BEAM_MODEL to a chat model ID from the catalog:

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["BEAM_TOKEN"],
    base_url="https://app.beam.cloud/v1",
)
response = client.chat.completions.create(
    model=os.environ["BEAM_MODEL"],
    messages=[{"role": "user", "content": "Explain what a container is in one sentence."}],
)
print(response.choices[0].message.content)

Pass stream=True for streaming responses. Supported options vary by model.

Other Model Types

Use /v1/embeddings for embeddings or /v1/models/{model-id}/invoke for models with custom schemas. The model’s catalog page includes request examples, image routes, and pricing. Inference requires a workspace token and credits for prepaid workspaces. Keep the token on your server. To deploy your own model, use an endpoint or container.