Scale Labs debuts new Refactoring Leaderboard

Scale Labs’ SWE Atlas Refactoring Leaderboard tests whether AI coding agents can restructure real codebases without changing behavior. Opus 4.7 leads, but results show reliability, cleanup, and repeatability still limit production use.

🗞 #ai @testingcatalog

Scale AI published SWE Atlas Refactoring Leaderboard, a new benchmark that evaluates agent capabilities of restructuring the code.

> It requires agents to produce twice as much lines of code than SWE Bench Pro.

> Claude Code with Opus 4.7 tops the leaderboard followed by Codex with GPT-5.5, GPT-5.4 and GPT-5.3.

> Refactoring is quite an important task for LLMs to handle as it often boils down to a quite boring engineering work.

OPENAI 🚨: 3 new models are now available on OpenAI Playground and APIs.

- gpt-realtime 2
- gpt-realtime-whisper
- gpt-realtime-translate

ChatGPT Voice Mode upgrade soon? 👀

GOOGLE 🚨: Gemini 3.1 Flash Lite is now Generally Available! Users can also test this model on AI Studio.

> Designed for ultra-low latency, high-volume tasks, and unmatched cost-efficiency, Flash-Lite is already transforming how applications are built at scale.

SPACEXAI 🚨: New signs of Grok Computer have been spotted on the Grok web.

A new selector allows users to choose between Grok Computer and a "Folder on Google Drive."

This feature became recently available to everyone and might not be intentional.

Grok Computer soon? 👀

OPENAI 🔥: Codex is getting a dedicated Chrome extension soon!

> With the new extension for Chrome, Codex is even better at working with apps and websites in your browser. It works in parallel across tabs in the background without taking over your browser, and you stay in control of which websites Codex can use.

* Not available yet 👀

Codex Chrome extension is now officially rolling out on macOS and Windows. You need to install the Chrome plugin to start testing.

SpaceXAI prepares Grok Build desktop app to rival OpenAI Codex

SpaceXAI appears close to launching Grok Build, a desktop coding app for macOS, Linux, and Windows.

> It will support planning mode, Plugins, Skills, and MCPs.
> Will be able to work with the Git tree, spawn dev servers, and work with a built-in browser.

🗞 #grok @testingcatalog

OpenAI launches new realtime voice and translation AI models

OpenAI added three real-time voice models to its API: GPT-Realtime-2 for complex voice agents, GPT-Realtime-Translate for multilingual speech, and GPT-Realtime-Whisper for live transcription, with pricing and Playground access.

🗞 #chatgpt @testingcatalog

Telegram ships major update for AI bots and automations

Telegram is slowly becoming the most AI-integrated messenger! Users can now interact with guest AI bots, build automated workflows involving bot-to-bot communication, and attach their personal AI assistants to their profiles so the bot can handle incoming inquiries.

🗞 #telegram @testingcatalog

Google unveils Google Health app, Health Coach, and Fitbit Air

Google expanded its health lineup with a unified Health App, AI-powered Health Coach under Premium, and the screenless Fitbit Air. The platform combines health data, supports secure sharing, and offers personalized wellness guidance.

🗞 #googleapps @testingcatalog

Google launches Gemini 3.1 Flash-Lite in General Availability

Google launched Gemini 3.1 Flash-Lite for Google Cloud, offering low-latency, high-volume AI with text and image support. Aimed at enterprise use, it delivers fast, cost-efficient performance for automation, tool calling, and real-time tasks.

🗞 #aistudio @testingcatalog

OpenAI adds Chrome plugin and tests Remote control for Codex

1. OpenAI expanded Codex with a Chrome extension that runs in an isolated browser session, supports background tab work, and uses DevTools.

2. OpenAI continues working on Remote Control for Codex, which will allow it to manage remote instances via SSH.

3. A new toggle has been added to the settings that will keep remote connection to Codex open.

🗞 #chatgpt @testingcatalog

Google is testing the option to mark Notebooks as "Donation Safe" as part of a Data Donation feature.

> Your logs from using this notebook will NOT be scrubbed (this allows for quality improvement).

> The notebook will immediately lose its Donation Safe status if shared.

> Marking a notebook as Donation Safe allows you to Donate Detailed Feedback.

> You must not donate NTK or Privileged data.

Hooks support is coming to the Codex app as well!

> Hooks are an extensibility framework for Codex. They allow you to inject your own scripts into the agentic loop.

Baidu released Ernie 5.1 preview which lands at rank 4 on Search Arena.

> Built on ERNIE 5.0's pre-training foundation, our latest foundation model upgrades search, reasoning, knowledge Q&A, creative writing, and agentic capabilities.

Benchmarks 👀