# d3

Multimodal decision model that reads each question in its own pass and returns a probability for every option.

Model id: `d3`
Context: 256K tokens
Quantization: bf16
Regions: ATL
Hugging Face: [vllm-sr/d3](https://huggingface.co/vllm-sr/d3)

## Pricing

- Tokens: $0.10 / 1M tokens

## Inputs

- text: context 256K tokens
- image: sources base64; formats png, jpeg; max size 16 MB; references type: integer, min: 1, max: 8

## Output: decisions

| Parameter | Type | Values | Default |
| --- | --- | --- | --- |
| `question_type` | string | noul, choice, score |  |

## Endpoints

- [`POST /v1/systemone`](/docs/reference/create-systemone.md): Create Decision