🚨 AI News | TestingCatalog

Latest AI News on AI Agents, Model Releases, Tools, Leaks, and Rumors 🗞️

[🚨 AI News | TestingCatalog @testingcatalog]

ANTHROPIC 🔥: A new Managed Projects feature has been spotted in testing on Claude.

> Claude takes on tasks and keeps the project organized.

> A project is a home for one stream of work. Sessions share memory and instructions so context carries forward, and Claude runs more autonomously. Projects only create and manage cloud sessions.

This feature may be built on top of Claude Managed Agents, where each project gets a dedicated cloud environment so Claude can execute periodic tasks and refine project context via "Dreams". It could also be a successor to Conway, which is set to be removed this Friday internally.

[🚨 AI News | TestingCatalog @testingcatalog]

Users can also add a prompt to shape what the model generates. The footage still dictates when each sound is added.

Video inputs can run up to three minutes, and the model pairs with Sonilo's Video2Music and Text2Music, meaning the same upload can generate sound effects and music.

Day zero API access ships through Fal as the exclusive launch partner.

Test it out 👀
https://t.co/H4z28UPgp5

[🚨 AI News | TestingCatalog @testingcatalog]

Sonilo has released Sound Effects 1.0, a video-native audio model that can read on-screen motion, scene context, and timing.

> It can make sound effects synced to the footage.
> Video-to-Sound-Effects runs automatically without requiring a prompt.
> Text-to-Sound-Effects is able to generate standalone assets from a written description.

[Sonilo @Sonilo_music]

Meet the first Sound World Model from Sonilo.

Generated video can look incredible, but without the right sound, it still feels distant.

Sonilo takes a video and creates the music and sound effects it needs, timed to the scene, motion, mood, and environment.

And in head-to-head benchmarks, Sonilo outperforms leading models across both music and sound effects.

With Sound Effects 1.0, we’re bringing music and sound effects together into one sound layer for videos.

Music carries emotion.
Sound effects create presence.

Together, they make worlds feel real, immersive, and alive!

Sound Effects 1.0 is live.

#SoundWorldModel #Sonilo #AIAudio

[🚨 AI News | TestingCatalog @testingcatalog]

@RobertJBye That’s a big upgrade! Congrats on the launch 🚀

[🚨 AI News | TestingCatalog @testingcatalog]

OPENAI 🔥: ChatGPT Voice is now rolling out on the desktop app to allow users to operate multiple agents across ChatGPT Work and Codex!

> New voice mode is powered by recently released GPT Live 1, a full-duplex voice model from OpenAI.

> Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

> ChatGPT Voice in Codex can also be used from the ChatGPT iOS app via paired remote access.

Android soon 👀

[OpenAI @OpenAI]

ChatGPT Voice is now in the desktop app.

Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice.

It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time.

Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

[🚨 AI News | TestingCatalog @testingcatalog]

@Spiderjin

[🚨 AI News | TestingCatalog @testingcatalog]

ANTHROPIC 🔥: Claude Voice Mode got upgraded with support for Opus 4.8 and Sonnet 5 models!

> Support for additional languages has been announced as well, including Spanish, French, Hindi, and Japanese!

> Claude Voice Mode can now take over complex tasks that require AI to use tools and Connectors.

> A new Voice Mode experience is rolling out on web and mobile apps.

As it was foretold 👀

[Claude @claudeai]

Voice conversations now use more of the models you have in chat, including Claude Opus and Sonnet. Claude can also reach the tools you've connected mid-conversation, like your email and calendar.

[🚨 AI News | TestingCatalog @testingcatalog]

@claudeai I really liked the upgrade; it is now capable of handling complex tasks!

Now it also needs proper support for Car Bluetooth - currently, it can't hear you properly unless you are speaking very, very loudly (I was testing it 8 hours ago or so)

[🚨 AI News | TestingCatalog @testingcatalog]

@JacobShintas It would look like this

[🚨 AI News | TestingCatalog @testingcatalog]

SPACEXAI 🔥: Grok 4.5 is now available on X, Grok web and mobile to all users.

It is also now possible to see the exact model name again on the model selector!

[Grok @grok]

Grok 4.5, our most capable model yet, is now available across https://t.co/f3u3RwhHjk, X, and the iOS and Android apps.

[🚨 AI News | TestingCatalog @testingcatalog]

@OpenAI @ajambrosino Niiice! Was waiting for it for a while!

Testing time! 🔥

[🚨 AI News | TestingCatalog @testingcatalog]

@DoctorYev @Uare_ai 👀👀👀

[🚨 AI News | TestingCatalog @testingcatalog]

MICROSOFT 🔥: New MAI-Image-2.5-Pro and MAI-Voice-2-Flash models have been announced!

MAI-Image-2.5-Pro is now in public preview. Pro is our highest-fidelity image model to date, priced at $5 per 1M text input tokens, $8 per 1M image input tokens, and $106 per 1M image output tokens.

MAI-Voice-2-Flash is now in public preview. Flash is 2x faster than MAI-Voice-2 and 32% cheaper, priced at $15 per 1M characters.

[Mustafa Suleyman @mustafasuleyman]

MAI-Voice-2-Flash launches today! Flash is 2x faster than MAI-Voice-2 and 32% cheaper, at $15 per 1M characters.

MAI-Voice-2-Flash is also in public preview and powers Dynamics 365 Contact Center, our enterprise platform for call center agents, and reduces GPU costs up to 89%.

More here...
https://t.co/JVU42BGuQ7

[🚨 AI News | TestingCatalog @testingcatalog]

The service runs on JMAP, allowing agents to work with a mailbox API, and proof of work plus a reputation score handling the rate limits instead of captchas or card checks.

The open alpha is free right now.

Test it out 👀
https://t.co/uF2CG0k7IS

[🚨 AI News | TestingCatalog @testingcatalog]

Atomic Mail put OpenClaw and Hermes under a test on top of an isolated email inbox for AI agents.

Both agents work through MCP or an Agent Skills. 

Test prompt: “Research both agents, debate with escalating fact-based emails, then honestly declare the winner.“ 

Who is the winner? 👀

[Atomic Mail @atomic_mail]

Atomic Mail put OpenClaw and Hermes head-to-head

We gave OpenClaw and Hermes one inbox each and the same prompt: prove you’re better. Who won?

Get your agent an inbox in 30 seconds:
– Plug in with one prompt via MCP or Agent Skill
– No CAPTCHA, no signup
– Run any workflow over email

Try the same prompt from the replies and share what happens.

[🚨 AI News | TestingCatalog @testingcatalog]

OpenAI started a broader rollout of ChatGPT Health for US users.

> ChatGPT can draw on your connected medical records and Apple Health information to help you see a more complete picture, track changes over time, and have more informed, personalized conversations.

Again 👀

[OpenAI @OpenAI]

Health in ChatGPT is starting to roll out to U.S. users.

You can securely connect Apple Health and supported medical records to understand your information in context, track what has changed, and have more informed conversations.

https://t.co/W2E6oT8c91

[🚨 AI News | TestingCatalog @testingcatalog]

Documented 🗞️
https://t.co/GHSTOxBFMs

[🚨 AI News | TestingCatalog @testingcatalog]

Offloop has launched a Channel-based workspace where multiple AI agents can work together on recurring tasks.

A custom dispatcher model called D1 decides which agent acts at each step.

> 84.9% on GDPval.
> 67.2% on JobBench.
> At around 1/5 of the cost frontier of multi-agent systems it tested against.

Agents with a workplace 👀

[Offloop @Offloop]

Introducing Offloop!

We're a team of four. Today our multi-agent harness hit state of the art on GDPval, ahead of Claude code and Codex across jobs that pay $2.4 trillion a year in the US.

Offloop gives every knowledge worker what the Fortune 500 spends billions on: a high-performing agent army that runs itself and grows the business.

[🚨 AI News | TestingCatalog @testingcatalog]

Channels keep context, files, schedules, and approvals together, allowing agents to pick up a task and carry it.

Users can connect to Notion, GitHub, GitLab, Sentry, Supabase, Vercel, Railway, Cursor, and Codex, while Teams can connect their own AI subscription or run on Offloop inference.

Invite code for the beta 🔑
https://t.co/jWoz3MBcmS

[🚨 AI News | TestingCatalog @testingcatalog]

Documented 🗞️
https://t.co/n8s3OoYjsk

[🚨 AI News | TestingCatalog @testingcatalog]

SPACEXAI 🔥: Grok Build will be able to deploy vibe-coded apps and serve them via a custom domain under grok[.]com!

> A new Build option will become available in the model selector where users will be able to ask Grok to vibecode games, dashboards, and other apps.

> The same experience has been in development on the Grok app for iOS as well.

Soon 🤖