# JigsawStack Updates: Meet OmiAI, AI Scraper Upgrades, Image translations, and More!

Hey Devs! 🎉

We have some amazing updates to share with you all! From the launch of **OmiAI** to powerful enhancements in our AI tools, we’re making AI development easier, smarter, and more efficient.

## What’s New?

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1740536584056/25251a3c-6adc-4956-911e-40ce3661fd6e.png align="center")

### Introducing OmiAI

Say hello to **OmiAI**, an opinionated AI SDK for **TypeScript** that **auto-picks the best model** for your prompts from a **curated suite** of LLMs. No more over-configuring—just **prompting**.

### Why OmiAI?

✅ **Curated model selection** for best quality, speed & cost  
✅ **Auto model chaining** for complex tasks  
✅ **Built-in reasoning** (o3-mini, DeepSeek r1)  
✅ **Internet access in real-time**  
✅ **Multi-modal support** (PDFs, images, audio, CSV, JSON, etc.)  
✅ **Seamless model fallbacks & retries**  
✅ **Powered by Vercel's AI SDK & JigsawStack tools**

**Quick Install:**

```json
npm i omiai
```

Learn more [**here**](https://jigsawstack.com/blog/introducing-omiai) & view models used [**here**](https://github.com/JigsawStack/omiai/blob/main/index.ts#L226)**.**

### Updated AI Web Scraper

Our AI Web Scraper just got an amazing update! Now, scraping and crawling the web is more seamless than ever:

✅ **AI Web Scraper now returns all links** on a page—perfect for web crawlers  
✅ **Autoscroll now on AI Web Scrapper** to ensure full-page coverage  
✅ **Increased** `goto` **option max to 40 seconds** for broader page coverage  
✅ `waitFor` **option max increased to 40 seconds** for better scraping precision

Check out the [**updated docs here**](https://jigsawstack.com/docs/api-reference/ai/scrape#body)**.**

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1740536618810/ef2759c3-7a77-4617-bbd4-a6c49f4b1bc7.png align="center")

### Image-to-Image Translation Model

We're working on an **Image-to-Image Translation Model** that extracts, translates, and overlays text back into images while preserving their original style. Our first version focuses on **OCR, translation, inpainting, and overlaying text**.

The long-term goal is to train a **diffusion model** for high-quality translations in multiple scripts, including Hindi, Arabic, and Chinese. We’re experimenting with different solutions and would love to hear your thoughts!

Learn more [**here**](https://jigsawstack.com/blog/diffusion-model-text-rendering).

### New SDK Install Option – Now Supports Bun!

Developers using **Bun** can now install JigsawStack with:

```bash
bun i jigsawstack
```

Join the conversation on our [**Discord**](https://discord.gg/GnvmdZ7z) or [**X**](https://x.com/jigsawstack) to let us know what Install Options you’d like to see next!

## Stay Ahead in 2025

We’re committed to **pushing AI boundaries** with tools that simplify workflows and boost efficiency. Expect more updates soon and thank you for being part of our journey!

Also, **We’re hiring!** If you’re a [**Full Stack Engineer**](https://yoeven.notion.site/Full-Stack-AI-Engineer-17595f7334d38049afddc936266846c7) join our mission to redefine AI development.

**Cheers,**  
Angel & The JigsawStack Team