Skip to content
Model infoModelsSnapshot: 15 Oct 2026

AI21 Labs / Task-specific models

Grammatical Error Correction

Specialized grammar-correction model distributed through AI21 Studio and a SageMaker model package. It returns suggested edits and their locations; the documented SageMaker input limit is 500 characters.

Text analysis
historicalClosed weightsText analysisGrammar correctionRewriting
Official documentation (opens in a new tab)Reviewed 14 Oct 2026
7 of 21 specification fields collected
Text input limit500500
LanguagesNot collectedAwaiting source review
Input billingNot collectedAmazon SageMaker (historical reference)
Output billingNot collectedNo inferred price

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

No exact-version evidence collected; specifications of another model are not inherited.

Maximum input tokens
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Maximum output
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Input modalities
TextSource for Input modalities (opens in a new tab)
Output modalities
Text correctionsSource for Output modalities (opens in a new tab)
Reasoning
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Tool calling
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Structured output
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Streaming
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Batch processing
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Fine-tuning
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Context caching
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Knowledge cutoff
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Release date
2023-03-09Source for Release date (opens in a new tab)

Wordtune task-API announcement, not the verified date of the SageMaker package.

Parameters
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Architecture
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Languages
Not collected

No exact-version evidence collected; specifications of another model are not inherited.

Maximum prompt characters
500Source for Maximum prompt characters (opens in a new tab)

Limit stated in the provider SageMaker notebook; a character count is not a token context window and is not assumed to apply to every later host.

Classification labels
Grammar, Missing Word, Punctuation, Spelling, Word Repetition, Wrong WordSource for Classification labels (opens in a new tab)
API request fields
text: the passage to correctSource for API request fields (opens in a new tab)
Response metadata
Response ID and corrections containing suggestion, start/end index, original text and correction type.Source for Response metadata (opens in a new tab)

Fixed service response fields do not establish arbitrary schema-constrained LLM generation.

Access & pricing

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

Amazon SageMaker (historical reference)

Provider source (opens in a new tab)
API model ID
gec
Access
Official SDK package lookup uses model_name="gec". Deployment creates a user-named endpoint. The surviving example does not establish current subscription availability.
Regions
Not collected
Inference instance types
ml.g4dn.2xlargeSource for Inference instance types (opens in a new tab)

Example instance only.

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

AI21 Studio (historical reference)

Provider source (opens in a new tab)
API model ID
Not collected

Service route is documented, but no separately selectable checkpoint ID is provided.

Access
Historical SDK POST route: /gec relative to the configured Studio API base URL. Current availability is unverified.
Regions
Not collected

Endpoint pricing has not been collected. Check the provider for applicable rates.

Data & enterprise terms

Retention, training use, regional processing and enterprise terms have not been summarized for this endpoint.

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.

Amazon SageMaker (historical reference)

gec
Not verified

No version-specific cache contract verified.

How it works
Not collected

No version-specific cache contract verified.

Minimum reusable input
Not collected

No version-specific cache contract verified.

Lifetime & refresh
Not collected

No version-specific cache contract verified.

How to enable it
Not collected

No version-specific cache contract verified.

Usage fields
Not collected

No version-specific cache contract verified.

Billing rules
Not collected

No version-specific cache contract verified.

Limits & data boundaries
Not collected

No version-specific cache contract verified.

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

Caching reviewed 14 Oct 2026

AI21 Studio (historical reference)

Not verified

No version-specific cache contract verified.

How it works
Not collected

No version-specific cache contract verified.

Minimum reusable input
Not collected

No version-specific cache contract verified.

Lifetime & refresh
Not collected

No version-specific cache contract verified.

How to enable it
Not collected

No version-specific cache contract verified.

Usage fields
Not collected

No version-specific cache contract verified.

Billing rules
Not collected

No version-specific cache contract verified.

Limits & data boundaries
Not collected

No version-specific cache contract verified.

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

Closed-weight distribution inference: the provider supplies a hosted API and a private SageMaker model package, not downloadable weights. The reviewed official AI21 public model inventory contains no matching Jurassic or GEC checkpoint. Public SDK source code is not model weights.

Classification evidence (opens in a new tab)

Sources reviewed