Moondream 3 Preview
Moondream 3 Preview is a large language model made by Moondream, added to the catalog on 15 Oct 2026. It has a context window of 32,000 tokens and costs $0.30 per 1M input tokens and $2.50 per 1M output tokens. It accepts text and images and produces text. Typical uses: answering questions about images, captioning images and extracting text from images with OCR.
Price checked ยท source: OpenRouter
What Moondream 3 Preview is
Moondream 3 Preview is a vision-language model from Moondream. It accepts text and images and produces text. Its listed tasks include visual question answering, image captioning, optical character recognition (OCR), and visual reasoning. This makes it suitable for prompts that ask for an explanation or description of visual content, as well as text extraction from images. The model is marked as supporting reasoning.
It has a context length of 32,000 tokens and a maximum output of 4,096 tokens. The context limit gives it room to consider a substantial amount of prompt content alongside visual input, while the output limit sets the maximum length of a single response. The catalog identifies it as a preview model and describes its architecture as a mixture of experts, with 9 billion total parameters and 2 billion active parameters.
These capabilities suit workflows centered on interpreting images through text. A prompt can ask questions about image content, request a caption, or seek visible text through OCR. Visual reasoning suits prompts that require drawing an answer from details in an image. The model's text output can present the response in prose. Its catalog lists text and images as inputs and text as output, so the described interaction centers on interpreting visual content and responding in text.
Key features
- Accepts text and images, and produces text
- Supports visual question answering, captioning, OCR, and visual reasoning
- Has a context length of 32,000 tokens
- Supports a maximum output of 4,096 tokens
- Cataloged as a mixture-of-experts model
Typical uses
- Answering questions about images
- Captioning images
- Extracting text from images with OCR
- Reasoning about visual details
Moondream 3 Preview pricing
US dollars, list price through OpenRouter.
| Item | Standard |
|---|---|
| Input, per 1M tokens | $0.30 |
| Output, per 1M tokens | $2.50 |
No price change since it was added on 15 Oct 2026.
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0009 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0095 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.014 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.049 |
Moondream 3 Preview specifications
| Model id | moondream/moondream3-preview |
|---|---|
| Version | moondream3-preview-20261013 |
| Context window | 32,000 tokensAbout 24,000 English words |
| Max output | 4,096 tokens |
| Knowledge cutoff | Not published |
| Added to catalog | 15 Oct 2026 |
| Input types | text, image |
|---|---|
| Output types | text |
| Tool calling | No |
| Structured outputs | No |
| Reasoning effort control | No |
| Content moderation | Not applied by provider |
Sample outputs
The same prompts are given to every model, and the unedited output is shown with the date it was produced. Upend does not publish leaderboard scores it did not run.
No output has been recorded for Moondream 3 Preview yet.
Compare Moondream 3 Preview
Side by side with other models: prices, context window, what each accepts and supports, and the same four tasks priced on both.