OrcaRouter, an OpenAI-compatible large language model gateway, has launched OrcaVerify Text 1.0, a text identification API designed to distinguish between human-written and AI-generated content. The system classifies submitted text into four categories and can identify potentially AI-generated material at the paragraph level rather than assigning only a single label to an entire document. The product supports Korean, English, Chinese, and Japanese and is aimed at organizations handling reviews, applications, manuscripts, registration information, and other content where verifying authorship may be important.
Four-Way Text Classification
OrcaVerify Text 1.0 categorizes content as HUMAN, AI_ASSISTED, AI_GENERATED, or ABSTAIN, giving users more nuanced results than a basic binary AI-or-human determination. Each response also includes an adjusted probability indicating how likely the content is to have been generated by AI, together with the decision threshold used to produce the classification. OrcaRouter said this approach is intended to make the model's judgment more transparent by showing how the probability relates to the final classification.
Addressing Uncertain Results
The ABSTAIN category is designed for situations where the system does not have enough information to make a sufficiently reliable determination. This can occur when submitted text is too short or falls outside the range for which the detection model has been calibrated, rather than forcing the system to return an uncertain prediction. In those cases, OrcaVerify provides a machine-readable explanation indicating why a definitive classification was not produced.
Paragraph-Level Detection
One of the platform's key capabilities is paragraph-level analysis, which is intended for documents combining human-written material with AI-generated or AI-assisted sections. Instead of treating an entire document as having a single origin, OrcaVerify can highlight individual paragraphs that appear more likely to contain AI-generated text. OrcaRouter positions the capability as useful for reviewing submitted manuscripts, checking marketplace reviews and registration data, and identifying specific sections of longer documents that may require further examination.
Growing Demand for AI Content Verification
The launch comes as AI-generated writing is increasingly appearing across professional, academic, commercial, and online environments, complicating traditional approaches to content verification. Manual review can become difficult when AI-generated passages are edited by humans or blended with independently written material, particularly when organizations process large volumes of submissions. Text detection systems are consequently being developed as screening tools that can help reviewers determine which content may warrant additional scrutiny rather than relying entirely on manual assessment.
OpenAI-Compatible Integration
OrcaRouter has designed OrcaVerify Text 1.0 to work through an API structure compatible with OpenAI's Chat Completions format, allowing developers to use existing client implementations with limited changes. Users can configure OrcaRouter as the API base, select the OrcaVerify Text 1.0 model, and place the text being analyzed in the final user message before receiving a structured response. The service supports both the chat completions and responses endpoints, reducing the need for developers already working with OpenAI-compatible systems to adopt a separate software development kit.
Pricing and Accessibility
OrcaVerify Text 1.0 is priced at $2 per one million input tokens, positioning the service for applications that may require analysis of large volumes of text. The multilingual support also broadens its potential use across organizations processing content in major Asian and international markets rather than focusing exclusively on English-language material. By combining probability estimates, explicit thresholds, paragraph-level identification, and an option to withhold uncertain judgments, OrcaRouter is emphasizing structured results that can be incorporated into automated review workflows.
OrcaRouter's launch of OrcaVerify Text 1.0 reflects growing demand for tools capable of assessing content as AI-generated writing becomes increasingly integrated into everyday digital communication. Its four-category classification system and paragraph-level detection are designed to provide users with more detailed information while avoiding definitive judgments when available evidence is insufficient. With multilingual support and OpenAI-compatible API integration, the company is positioning the service as a scalable verification layer for platforms and organizations that need to examine large quantities of submitted text.