Extract
Extract is a large language model made by ScaleDown, added to the catalog on 13 Oct 2026. It has a context window of 1,000,000 tokens and costs $0.04 per 1M input tokens and $0 per 1M output tokens. It accepts text, images and files and produces text. Typical uses: extracting entities using custom field descriptions, pulling requested fields from text and extracting information from images.
Price checked ยท source: OpenRouter
What Extract is
ScaleDown: Extract is a description-driven entity extraction model from ScaleDown. Instead of relying on a fixed set of entity types, it lets callers describe the fields they want in plain English. Those field descriptions are supplied through the `provider.options.scaledown.entities` option. This makes the requested fields configurable for a task rather than tying extraction to a predefined list. The model accepts text, images, and files, and produces text. The facts describe its extraction approach but do not specify particular entity types or a fixed output schema.
The model has a context window of 1,000,000 tokens and a maximum output of 900,000 tokens. Its large stated context capacity may suit extraction tasks involving lengthy input, while its text output can present the information it extracts. The available details do not specify how output fields are formatted, or provide additional controls for shaping the response. Its central use is extracting information according to fields that the caller describes in plain English. Since it accepts text, images, and files, those are the listed input types for supplying material to the model. The facts do not provide further detail about handling each type, so they do not establish modality-specific extraction features. ScaleDown: Extract is suited to tasks where the requested fields need to be described for the extraction job, rather than selected from a stated fixed set of entity types. The listed context and output limits indicate the maximum token capacities; they do not describe a guaranteed result size for any particular request.
Key features
- Description-driven entity extraction with fields specified in plain English
- Accepts text, images, and files
- Produces text
- Has a 1,000,000-token context window
- Maximum output of 900,000 tokens
Typical uses
- Extracting entities using custom field descriptions
- Pulling requested fields from text
- Extracting information from images
- Extracting information from files
Extract pricing
US dollars, list price through OpenRouter.
| Item | Standard |
|---|---|
| Input, per 1M tokens | $0.04 |
| Output, per 1M tokens | $0 |
No price change since it was added on 13 Oct 2026.
| What a task costs | Cost |
|---|---|
| Answer a short question500 in / 300 out | $0.0000 |
| Summarise a 50-page PDF25,000 in / 800 out | $0.0010 |
| Review a 2,000-line source file30,000 in / 2,000 out | $0.0012 |
| Read a 300-page contract set150,000 in / 1,500 out | $0.0060 |
Extract specifications
| Model id | scaledown/extract |
|---|---|
| Version | extract-20260928 |
| Context window | 1,000,000 tokensAbout 750,000 English words |
| Max output | 900,000 tokens |
| Knowledge cutoff | Not published |
| Added to catalog | 13 Oct 2026 |
| Input types | text, image, file |
|---|---|
| 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 Extract yet.
Compare Extract
Side by side with other models: prices, context window, what each accepts and supports, and the same four tasks priced on both.