ApyHub is an AI-powered API platform that provides flexible pricing plans for developers and teams. Its artificial intelligence API catalog brings together a variety of hosted models, enabling tasks such as classification, summarization, sentiment analysis, content moderation, and information extraction through HTTP calls, so developers do not need to rent GPUs themselves, choose model weights, or maintain inference servers. A single API key covers all models in the catalog, billing is per call, and a free tier is available.
Main capabilities include:
- Resume Parsing API: Parses resumes and CVs into JSON, extracting contact information, work experience, education, and skills, supports PDF, Word, and scanned documents, covers more than 80 languages, and offers a free tier.
- Language Detection and Text Annotation API: Detects the language of any text and provides a confidence score, supports stemming and part-of-speech tagging, and offers a simple REST API and JSON responses.
- Image Foreground Mask Generation API: Separates the subject from an image and returns a mask or a background-removed image. Supports passing in a public image URL or base64-encoded image data, with the option to return JSON metadata or image bytes. The subject mask endpoint returns a grayscale cutout, suitable for compositing, edge cleanup, or downstream image analysis; the response includes the output format, media type, width and height, byte size, data URI, subject bounding box, foreground ratio, and the quality setting used, and the bbox is null when no subject is found. The background removal endpoint returns a ready-to-use subject image, supports cropping to the subject bounding box, adding padding, limiting the output size, and optionally best or fast quality. The output format can be json, png, webp, or jpeg; JPEG requires an opaque background color in #RRGGBB form. Both endpoints accept images up to 20 MB and return a data URI and image metadata, making it easy to integrate directly into storage, preview, or rendering workflows.
- Food Search and Q&A API: Searches food content, classifies food images, and gets food chatbot replies, returning recipes, videos, Q&A, jokes, and structured search results.
- News API: Searches, extracts, and retrieves news articles filtered by URL, ID, or source, and gets titles, authors, summaries, publication dates, sentiment, and links.
- Qwen3-Next-80B Chat Completion: Generates AI chat replies through simple API integration, for creating chatbots, answering questions, and generating content.
- LTX 2 Video Generation (RTX 5090):Turn text prompts or source photos into short LTX-2 videos, running on a dedicated NVIDIA RTX 5090, and return a task ID to track progress. Text-to-video uses the POST /v1/video/generate endpoint, where you can specify seed, model, length, quality, and resolution; length supports 3s, 5s, and 10s, quality can be fast, balanced, or best, and resolution can be sd or hd. Image-to-video uses the POST /v1/video/i2v-ltx2 endpoint, sending a base64-encoded image and a prompt describing motion, where you can specify seed, length, quality, and resolution, bringing the photo to life while preserving the source image's composition. Both endpoints return status, message, and taskid; you can store the task and then poll the shared GET /v1/video/tasks/:taskid endpoint until the task completes or errors out. The status response includes type, status, taskid, and elapseds, and after generation completes it includes the output_url of the final MP4; on failure, you can check the error field in the task response. The service runs on LTX-2 19B (distilled version) on a dedicated NVIDIA RTX 5090, and both text-to-video and image-to-video take about 30-60 seconds to generate a clip of about 5 seconds, suitable for ad mockups, concept previews, social clips, and internal creative iteration.
- LTX Video video generation (RTX 5090): Use LTX-Video 13B on a dedicated RTX 5090 to generate short MP4 clips from text prompts or bring photos to life, asynchronous tasks, clip duration 4 to 5 seconds.
- Text-to-image task API: Use Flux.1-Schnell or SDXL-Turbo to generate images from text prompts, polling task status to get progress, errors, and the final output URL.
- AI-generated image detection API: Classify public JPG, PNG, or WebP URLs as likely AI-generated, human-made, or uncertain, returning ai_score, confidence, and source breakdown.
- Text similarity search API: Compare a source string and a target string, returning a similarity score from 0 to 1, suitable for fuzzy matching, duplicate checking, and text ranking.
- Market trends and growth forecasting API: Analyze a website's market trends, opportunity score, differentiation, and growth potential, suitable for product research and competitive review.
In terms of usage, developers call each API endpoint via HTTP, authenticate with a single API key, are billed by number of calls, and can start with a free tier. Some asynchronous tasks (such as video and image generation) require first creating a task, then polling the task status endpoint to get the result.
Applicable scenarios include: application development that needs to quickly integrate AI capabilities, resume screening, content moderation, image processing and background removal, news aggregation, chatbots, video and image generation, AI-generated content detection, text matching, and market research, among others.
In terms of limitations: the image foreground mask API accepts images up to 20 MB, and JPEG output requires an opaque background color; LTX 2 video generation outputs clips of about 5 seconds, with generation time of about 30-60 seconds; LTX Video outputs clips of 4 to 5 seconds; the specific quotas, rate limits, and pricing details for some APIs are not yet provided.
