> For the complete documentation index, see [llms.txt](https://docs.takionapi.tech/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.takionapi.tech/modules/ocr.md).
# Image-to-Text Captchas / OCR
## Why a custom model wins
An OCR model is only as good as what it was trained on. The public solvers see a huge generic mix and do fine on the average captcha. Your target isn't the average captcha.
Train on the specific font, warp, colors, and noise of one challenge and the accuracy jumps. It's not magic, it's just the right training set. A model that only has to read one style of captcha beats a model that has to read everything.
Other than that, using a big vision model that will use unecessary usage and memory on production for a small task, is never a good choice.
#### We ensure
* A near 100% pass rate
* A fraction of milliseconds for the resolution
* An almost inexistent resources usage
### **Example of Captchas**
Our OCR model accepts a wide range of images for solving and is optimized for specific captchas. We can also implement new captchas based on client requests with near-perfect accuracy.
{% hint style="info" %}
Facing those or other custom-website's captchas? [Contact us](https://takionapi.tech/contact) for finding the best solution for your use case
{% endhint %}
#### **Queue-IT's BotDetect**
Queue-IT and SoftBlock provide image-to-text challenges known as BotDetect, which our API can solve efficiently.
#### **TicketMaster SG's Captchas (ATC & Checkout)**
Our API quickly and accurately handles TicketMaster SG's image-to-text captchas during ATC and Checkout processes, using advanced pre- and post-processing of images.