LanguageCheck API
Version: 2.0.0
Base URL: https://api.languagecheck.ai/v2
Overview
The LanguageCheck API provides programmatic access to the services offered by languagecheck.ai. It exposes a RESTful interface over HTTPS and is designed for integration into translation workflows, quality assurance pipelines, and any application that needs automated assessment of bilingual content.
Endpoints
GET /status — Service Health Check
This endpoint returns the current operational status of the LanguageCheck service and its underlying AI infrastructure. It accepts no parameters.
The response indicates whether the platform and the AI service are fully operational, experiencing degraded performance, or undergoing maintenance. It is useful for health checks in monitoring dashboards, or for verifying service availability before submitting translation segments.
GET /wallet — Remaining Words Balance
This endpoint returns the current number of words still available for the authenticated user. It is useful for checking quota before sending translation jobs and for surfacing remaining capacity in your application.
The response includes the remaining word count and the timestamp of the last wallet update, which can be used to keep UI indicators or internal monitoring in sync.
POST /check — Translation Quality Evaluation
This is the primary endpoint of the API. It accepts a bilingual segment — a source text and its corresponding translation — along with the respective language codes, and returns a structured evaluation of the translation quality.
The evaluation covers three dimensions: major errors (issues that distort meaning or render the translation unsuitable), minor errors (inaccuracies or stylistic imprecisions that do not significantly affect usability), and fluency issues (problems with the naturalness and readability of the target text, independently of its accuracy). Each dimension is accompanied by a textual explanation when issues are detected.
The response also includes an overall verdict for the segment and an ambiguity flag, which signals that the model could not assess the segment with sufficient confidence and that human review may be needed.
Explanations can be returned in a configurable language, making it straightforward to surface QA feedback directly to translators or reviewers in their preferred language.