Extract text from images with Muapi OCR Recognize Text. Fast local OCR returns each fragment's location and confidence. Try free, pay per generation. Detect and extract text fragments and their positions from an image using local OCR.
Detect and extract text fragments and their positions from an image using local OCR.
About this model
Muapi OCR Recognize Text detects and extracts every readable text fragment inside an image, returning each fragment's string content, its bounding box as a fraction of the image dimensions, and a confidence rating (high, medium, or low). It runs on a dedicated local-OCR pipeline rather than a general vision-LLM call, which keeps recognition fast and cheap for high-volume workloads.
This makes it a natural first step before an image-editing pass: detect what text exists and where, let a user or downstream tool decide what to change, then feed the edited fragments back into an image-edit model like GPT Image 2 or Nano Banana Pro to regenerate the image with only the targeted text replaced.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.02 per generation | Flat rate per image, regardless of the number of text fragments detected. |
| Fal.ai | Not available | No dedicated local-OCR text-detection endpoint listed. |
| Replicate | Varies by model | OCR models exist individually but are not bundled into a bbox+confidence text-detection API like this one. |
Flat rate per image, regardless of the number of text fragments detected.
No dedicated local-OCR text-detection endpoint listed.
OCR models exist individually but are not bundled into a bbox+confidence text-detection API like this one.
** Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Image URL | string | The URL of the image to recognize text from. | https://d3adwkbyhxyrtq.cloudfront.net/ai-images/example-poster.jpg |
The URL of the image to recognize text from.
https://d3adwkbyhxyrtq.cloudfront.net/ai-images/example-poster.jpgDeveloper documentation
Prepare your image: Have a public image URL ready — a poster, screenshot, product photo, or any image containing visible text.
Call the endpoint: Send a POST request with the image_url field. No other parameters are required.
Read the results: The response's items array lists each detected text fragment with its text string, bbox (as [x, y, w, h] fractions of image width/height), and a confidence label (high, medium, or low).
Chain into an edit (optional): Use the fragment list to build an edit prompt — e.g. "replace the fragment at this bbox with new text, keep everything else unchanged" — and pass it to an image-edit model.
curl -X POST https://api.muapi.ai/api/v1/ocr-recognize-text \
-H "Content-Type: application/json" \
-H "x-api-key: YOUR_API_KEY" \
-d '{
"image_url": "https://example.com/poster.jpg"
}'
Frequently asked
It scans an input image and returns every detected text fragment along with its bounding box (as fractions of image size) and a confidence score, without regenerating or modifying the image.
The only required input is `image_url`, the public URL of the image to scan. An optional `webhook_url` can be supplied to receive the result asynchronously.
Fractional coordinates (0–1 range) let the same bounding box be applied correctly regardless of any resizing that happens between detection and a later edit step, without extra unit-conversion math.
Each fragment is labeled `high`, `medium`, or `low` based on the OCR engine's internal recognition score, so downstream logic can choose to ignore low-confidence noise.
OCR Recognize Text only detects and returns text fragments. To change the text, pass the detected fragments and your desired replacements to an image-edit model as a follow-up call.
Standard web image formats (JPG, PNG, WebP) served from a publicly reachable URL are supported.