DEEPDENSE / DEVELOPERS

Less setup.
More understanding.

Explore the request format, choose a preview model,
and see how your workflow could fit together.

API PLAYGROUND

Shape the request.
Inspect the format.

Choose a model, set system instructions, add a prompt, and adjust temperature. The request preview updates as you edit.

Running a request returns “Not enough credits.” No inference provider is called and no model output is generated.

REQUEST FORMAT PREVIEW

{
  "model": "deepdense-extract",
  "messages": [
    {
      "role": "system",
      "content": "Extract only the fields present in the document."
    },
    {
      "role": "user",
      "content": "Your document text here"
    }
  ],
  "temperature": 0.2
}

Model catalogue

Browse proposed models for general assistance, structured extraction, and classification. Each entry is labeled as a preview.

Explore models

API keys

Explore the key management screen. Creation is blocked at zero credits, so no usable credentials are generated.

View API keys

Integration planning

Use the visible request format to plan your workflow. A public inference API and installable SDK are not available yet.

Choose a use case

COMMON QUESTIONS

A few things to know.

Can I run inference today?

No. Playground execution is blocked with “Not enough credits.” The catalogue contains preview entries rather than deployed models.

Can I create a production API key?

No. Key creation is credits-gated and there is no credit purchase flow in this release.

Can I fine-tune a model?

Training provider connections are not enabled. Experiment configuration is part of the product design, with creation currently credits-gated.

What does the evaluation workflow measure?

The workspace supports exact match and token overlap for expected and actual outputs. These measures do not establish semantic correctness. New evaluations are currently credits-gated.

Less setup.More understanding.

Shape the request.Inspect the format.