Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Upstage / Document Classify

Document Classify (260304)

Document Classify versioned service. Updates Japanese handwriting and table recognition.

Document extraction / OCR
availableClosed-source managed componentDocument understanding
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
12 of 24 specification fields collected
Context windowNot collectedAwaiting source review
Maximum outputNot collectedAwaiting source review
Input billingSee ratesUpstage API
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Maximum input tokens
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Maximum output
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Parameters
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Architecture
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Languages
Alphanumeric, Hangul, Hiragana, Katakana and Hanja supported; Hanzi and Kanji are limited/beta.Source for Languages (opens in a new tab)

OCR character coverage is not a claim of unrestricted language understanding.

Knowledge cutoff
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Release date
2026-03-04Source for Release date (opens in a new tab)
Reasoning
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Tool calling
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Structured output
SupportedSource for Structured output (opens in a new tab)

User-defined class labels supplied using JSON schema.

Streaming
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Batch processing
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Fine-tuning
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Context caching
Not collected

Not verified for this exact version. No current-family token limits, rates or capabilities inherited into historical releases.

Input modalities
Image, DocumentSource for Input modalities (opens in a new tab)
Output modalities
TextSource for Output modalities (opens in a new tab)
Release identity
document-classify-260304; current alias document-classifySource for Release identity (opens in a new tab)

Provider-listed service/model version; aliases are not counted as additional independent releases.

Version changes
Updates Japanese handwriting and table recognition.Source for Version changes (opens in a new tab)
Input file formats
JPEG, PNG, BMP, TIFF, HEIC, PDF, DOCX, PPTX, XLSX, HWP, HWPXSource for Input file formats (opens in a new tab)
Document limits
50 MB / 100 pages synchronously; 200 million pixels per page.Source for Document limits (opens in a new tab)

Pixel counts for non-image files are measured after conversion at 150 DPI. These current service limits are not copied to older versions.

Request limits
Only the first 100 pages are processed by the synchronous API when a file exceeds the limit.Source for Request limits (opens in a new tab)
Output formats
Classification label, Confidence metadataSource for Output formats (opens in a new tab)
Document extraction features
Choose the best matching class, optionally split mixed documents by type, and optionally supply additional split criteria.Source for Document extraction features (opens in a new tab)

Split is beta and defaults to false. Metadata tool_calls are not arbitrary tool execution.

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Upstage API

Provider source (opens in a new tab)
API model ID
document-classify
Access
API key/account required. Marketplace and private deployment contracts are separate.
Regions
Not collected

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Document classificationClassification1$0.004page

Rate conditions

  1. Direct API per processed page; VAT excluded. Standard/enhanced modes and Studio workflows have different contracts. No historical or marketplace rate inferred.
Pricing source (opens in a new tab)

Data & enterprise terms

Provider states input data is not used for training and is not stored except when required for service delivery and explicitly stated.

Read the service policy (opens in a new tab)

Upstage API

Provider source (opens in a new tab)
API model ID
document-classify-260304

Pinned version behind the documented current alias.

Access
API key/account required. Marketplace and private deployment contracts are separate.
Regions
Not collected

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 14 Oct 2026
CategoryUsageRateUnit
Document classificationClassification1$0.004page

Rate conditions

  1. Direct API per processed page; VAT excluded. Standard/enhanced modes and Studio workflows have different contracts. No historical or marketplace rate inferred.
Pricing source (opens in a new tab)

Data & enterprise terms

Provider states input data is not used for training and is not stored except when required for service delivery and explicitly stated.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Upstage API

document-classify
Not verified

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

How it works
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Minimum reusable input
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Lifetime & refresh
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

How to enable it
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Usage fields
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Billing rules
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Limits & data boundaries
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 14 Oct 2026

Upstage API

document-classify-260304
Not verified

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

How it works
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Minimum reusable input
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Lifetime & refresh
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

How to enable it
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Usage fields
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Billing rules
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

Limits & data boundaries
Not collected

No verified caching contract for this exact document model or service. No text-token cache behavior inferred.

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Caching reviewed 14 Oct 2026

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

This record represents a named document-processing API/service version, potentially combining detection, recognition, parsing and postprocessing. It does not establish an independent trained checkpoint for every release. Current OCR explicitly shares the Document Parse engine.

Classification evidence (opens in a new tab)

Lifecycle dates

Provider-listed release4 Mar 2026Source (opens in a new tab)

Sources reviewed