Name: Moondream 3 Preview API
Brand: Moondream

Overview

What Is Moondream 3 Preview?

Moondream 3 Preview is a vision-language model from Moondream. It is a mixture-of-experts model with 9B total parameters, of which 2B are active for each token, which keeps inference close to the cost of a small dense model. It reads images together with a text prompt and answers in text, covering visual question answering, captioning, OCR and visual reasoning. It is available on AI/ML API as moondream/moondream3-preview.

Moondream 3 Preview is the first release of Moondream's third generation. Its successor, Moondream 3.1, keeps the same 9B/2B layout, limits and price on AI/ML API.

Specifications

Developer: Moondream
Available on AI/ML API since: 15 October 2026
Architecture: mixture of experts, 9B total / 2B active parameters
Context window: 32,000 tokens
Max output: 4,096 tokens
Input: text and images — output: text
Capabilities: vision, streaming

What It Is Built For

The model targets high-volume image understanding rather than long conversations: answering a question about a photo or screenshot, writing a caption, reading the text in an image and reasoning about what the image shows. The 32,000-token context and 4,096-token output limit fit short, image-centred requests.

Pricing on AI/ML API

Input is billed at $0.41262 per 1M tokens and output at $3.4385 per 1M tokens.

Use cases

Visual question answering over photos, screenshots and product images, where an application asks a specific question and needs a short text answer.

Captioning and alt text for image libraries, catalogues and user uploads.

OCR: reading text from signs, labels, receipts, documents and UI screenshots.

Visual reasoning steps inside a larger pipeline, such as checking what an image contains before routing it to another model or tool.

How to use Moondream 3 Preview API

Install any OpenAI-compatible SDK, point it at api.aimlapi.com/v1, and set the model to moondream/moondream3-preview.

import requests

r = requests.post(
    "https://api.aimlapi.com/v1/chat/completions",
    headers={"Authorization": "Bearer " + AIMLAPI_KEY},
    json={
      "model": "moondream/moondream3-preview",
      "messages": [
        {
          "role": "user",
          "content": "Hello!"
        }
      ]
    },
)
print(r.json())

const r = await fetch("https://api.aimlapi.com/v1/chat/completions", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.AIMLAPI_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model": "moondream/moondream3-preview",
    "messages": [
      {
        "role": "user",
        "content": "Hello!"
      }
    ]
  }),
});
console.log(await r.json());

curl -X POST https://api.aimlapi.com/v1/chat/completions \
  -H "Authorization: Bearer $AIMLAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"moondream/moondream3-preview","messages":[{"role":"user","content":"Hello!"}]}'

OpenAI-compatible — swap the base URL and it works with your existing SDK.