[Dev Catch Up #55] - Gemma 3, OpenAI's new Response API, AI Agent SDK, Nuxt 3.16, OpenManus, Bun v1.2.5, CocoIndex, Copacetic, Zod, R1-Omni, Mox and much more.

Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles.

Welcome to the 55th edition of DevShorts, Dev Catch Up!

For those who joined recently or are reading Dev Catch Up for the first time, I write about developer stories and open source, partly based on my work and experience interacting with people all over the globe.

Thanks for reading Dev Shorts! Subscribe for free to receive new posts and support my work.

Some recent issues from Dev Catch up:

• How to build you own MCP Server?

• LLM Glossary

• How to use Postgres as a Vector Database with BGE Embedding model

• How to do Local AI Code Generation and Completion using Open-source Llama.cpp?

Join 6500+ developers to hear stories from Open source and technology.

Must Read

• Google has launched Gemma 3, which is already making waves with its massive 128k context window and support for over 140 languages. In other news Gemini 2.0 Flash (Image Generation) is now open to everyone. It excels at combining text with images and even lets you edit images via text prompts. You can access both models through AI Studio.

• OpenAI has rolled out two big updates, built-in tools in the Responses API covering web search, file search, and computer use, and a new AI Agent SDK to orchestrate agentic workflows. These additions will help developers on building agents. Check out this OpenAI Post to learn more on this.

• We all know Manus is a private AI agent, but now there is OpenManus, an open-source framework that lets you build AI agents without restrictions. Developed by contributors from MetaGPT, and this is easy to set up. If you want to create LLM based agents, head to their GitHub repo for all the details.

• In our last edition, we mentioned the TypeScript 5.8 release, and now there is another big update from Microsoft, that they are porting TypeScript compiler from JavaScript to Go to achieve a 10× speed boost with better concurrency and lower memory usage. The new version is expected by end of 2025. Check out the detailed blog to know why Go, why not C#, and more details.

OSS Highlight of the Week

CocoIndex is an open-source data indexing platform built for AI applications. In a typical RAG pipeline, indexing, embedding, and chunking can be time consuming. CocoIndex simplifies this by handling all your data related tasks like connecting data sources, choosing the right indexing strategy, chunking and embedding. It then provides a standard API for accessing index. That way, you can focus on building AI solutions rather than dealing with data tasks. Check out the GitHub repo for more details.

Good to know

• While exploring data validation in TypeScript, I came across Zod, a library that lets you define your data, validate it, and trust the types, all in one go. This blog covers everything about this library. If you need reliable data for forms or API responses, Zod can make your life easier.

• I read an interesting post about JavaScript usage on the web. The report highlights a continuous increase in JavaScript payloads, with median sizes reaching 558 KB on mobile and 613 KB on desktop. Read More information in this article.

• Monitoring internet traffic helps spot security threats, bandwidth hogs, and misbehaving apps and that’s where Sniffet comes in. This tool presents network activity in a simple interface, making it easy to detect unusual connections. For more details, see the project repository.

• I came across this interesting framework for testing AI agents. Factorio Learning Environment (FLE) tests AI by giving them tasks with fixed resources. If this interests you, check their GitHub repo for more details.

• All of us use LLMs for coding, but not everyone gets the best results. I got to read some blogs which share great insights on how to use LLMs more effectively for writing code. Read Harper’s blog and Simon Willison's blog to know how you can get the most out of LLMs.

• I used svg2pdf.js for browser-based SVG-to-PDF conversion in my recent project. This JavaScript library handles everything without needing a server. Easy to install via npm/yarn and perfect for apps needing downloadable reports or graphics. Check GitHub repo for more details.

• While exploring ways to simplify email management, I found Mox. It is a modern, secure, all-in-one server that keeps communication decentralized. Developed in Go, easy to set up and maintenance and open source as well. Check this page for more details on Mox.

Notable FYIs

• If you are a DevOps engineer, this might interest you. Copacetic is a Go-based CLI tool that patches container images on the fly, no full rebuilds needed. It is easy to install and only applies the necessary patches, saving time and space. Explore the source on GitHub for more details.

• As a developer, I work with multiple frameworks, and Nuxt.js is one of them. The latest Nuxt 3.16 release comes with the create-nuxt tool for starting projects, along with several new features. Read their official blog post for more details.

• Alibaba Cloud has already contributed the opensource community with Qwen Model, now they have introduced R1-Omni a RLVR-Powered Multimodal LLM that excels at emotion recognition. Read full details about this on Hugging Face page.

• Each week our newsletter covers interesting database developments, and this time it's PGlite, a tiny 3MB WASM version of Postgres that runs directly in browsers without extra dependencies. Perfect for projects needing database capabilities without infrastructure headaches. Find more details about PGlite in this GitHub repo.

That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable. A subscription to the newsletter will be awesome if you are reading it for the first time.