https://blog.google/technology/developers/gemini-code-assist-free/

Google launched a free version of Gemini Code Assist globally. It comes with:
- 180K code completions per month
- Support for all programming languages in the public domain
- 128K token context window

Material Theme has been pulled from VS Code's marketplace

https://web.archive.org/web/20250226020241/https://github.com/material-theme/vsc-material-theme/discussions/1313


"A member of the community did a deep security analysis of the extension and found multiple red flags that indicate malicious intent and reported this to us.

Our security researchers at Microsoft confirmed this claims and found additional suspicious code.

We banned the publisher from the VS Marketplace and removed all of their extensions and uninstalled from all VS Code instances that have this extension running.

For clarity - the removal had nothing to do about copyright/licenses, only about potential malicious intent."

https://openai.com/index/introducing-gpt-4-5/

openAI introduced GPT 4.5. GPT‑4.5 is a step forward in scaling up pre-training and post-training. By scaling unsupervised learning, GPT‑4.5 improves its ability to recognize patterns, draw connections, and generate creative insights without reasoning.

Some kind of bad news for Firefox users


// Update

https://blog.mozilla.org/en/products/firefox/update-on-terms-of-use/

https://kibty.town/blog/todesktop/

Interesting writeup about a vulnerability in the update mechanism for cursor, linear and other big apps which could be used to ship malicious updates.

https://github.com/qixucen/atom

https://arxiv.org/abs/2502.12018

Reasoning models lack atomic thoughts -unlike humans using independent units, they store full histories.

Atom of Thoughts (AOT): lifts gpt-4o-mini to 80.6% F1 on HotpotQA, surpassing o3-mini and DeepSeek-R1

Really interesting idea to dramatically reduce token size and improve model intelligence.

https://press.opera.com/2025/03/03/opera-browser-operator-ai-agentics/

Opera is adding an AI agent into the browser, making it the first major browser with AI-based agentic browsing.

https://devblogs.microsoft.com/typescript/typescript-native-port/

Microsoft released an official native port of the TypeScript compiler and tooling. The native implementation will drastically improve editor startup, reduce most build times by 10x, and substantially reduce memory usage.

Written in Go and available here:

https://github.com/microsoft/typescript-go


"…using the Visual Studio Code codebase as a benchmark, the current time to load the entire project in the editor on a fast computer is about 9.6 seconds. This drops down to about 1.2 seconds with the native language service, an 8x improvement in project load time in editor scenarios."

https://www.docker.com/llm/

Docker is preparing an alternative to e.g. Ollama which you can use to run LLM‘s.

Probably something like docker run model mistral/mistral-small

https://documents.craft.me/announcing-chaps

Chaps is a new tool created by Craft:

"In a world drowning in apps, notifications, and complexity – simplicity is the ultimate luxury. Meet Chaps, your personalized AI-powered agents that simplify life, streamline tasks, and put you back in control."

https://github.com/antiwork/gumroad

Gumroad is now Open Source!

https://firebase.studio

Google just released firebase.studio.
It’s their answer to cursor, windsurf, replit etc.

https://github.com/nari-labs/dia

1.5 researchers published a new open source TTS model which seems better than the current state of ElevenLabs (leading TTS company).

Check the demo page:

https://yummy-fir-7a4.notion.site/dia

https://blog.atuin.sh/atuin-desktop-runbooks-that-run/

A local-first, executable runbook editor for real terminal workflows.
Atuin Desktop looks like a doc, but runs like your terminal. Script blocks, embedded terminals, database clients and prometheus charts - all in one place.

https://blog.jetbrains.com/ai/2025/04/mellum-goes-open-source-a-purpose-built-llm-for-developers-now-on-hugging-face/

Jetbrains published an open source LLM focused on code completion.

https://huggingface.co/nvidia/parakeet-tdt-0.6b-v2

parakeet-tdt-0.6b-v2 is a 600-million-parameter automatic speech recognition (ASR) model by nvidia designed for high-quality English transcription, featuring support for punctuation, capitalization, and accurate timestamp prediction.

It seems better than whisper-3 large (https://huggingface.co/spaces/hf-audio/open_asr_leaderboard) with 1B less parameters.

https://www.reuters.com/business/openai-agrees-buy-windsurf-about-3-billion-bloomberg-news-reports-2025-05-06/

OpenAI has agreed to buy Windsurf, an AI coding tool, for about $3 billion, marking the ChatGPT maker’s largest acquisition to date.

https://github.com/colanode/colanode

Open-source & local-first Slack and Notion alternative.
Colanode is an all-in-one platform for easy collaboration that you can self-host and fully control.

https://github.com/bytedance/deer-flow

DeerFlow from Bytedance is a community-driven framework for deep research, combining language models with tools like web search, crawling, and Python execution.

https://github.com/astral-sh/ty

An extremely fast Python type checker and language server, written in Rust.

In early testing, it's 10x, 50x, even 100x faster than existing type checkers. ( >600x speed-ups over Mypy in some real-world projects.)