BREAKING 🔥: OPEANAI LAUNCHED GPT LIVE 1, A NEW FULL DUPLEX VOICE MODEL FOR CHATGPT!

It comes in 2 versions: GPT‑Live‑1 and GPT‑Live‑1 mini, and is rolling out to everyone in the coming days! API support is coming soon too.

> GPT‑Live is built on a full-duplex architecture, meaning it can listen and speak at the same time. During conversations, GPT‑Live can show it’s paying attention with phrases like “mhmm” or “yeah”, engage in quick back-and-forth, or just stay quiet when you need a moment to think.

> GPT‑Live is also our smartest voice model yet. For questions that require web search, deeper reasoning, or more complex work, it delegates to our latest frontier model behind the scenes and brings the result back into the conversation when it’s ready.

BREAKING 🔥: GROK 4.5 IS ROLLING OUT ON GROK BUILD, APIS AND XAI CONSOLE!

> Context window - 500K tokens
> Modalities - Text, Image → Text
> Pricing - Input $2.00 / 1M, Output $6.00 / 1M

V9 is finally here 🤖

SPACEXAI 🔥: Grok 4.5 performs on par with GPT 5.5 and close to Opus 4.8 on coding benchmarks.

> Grok 4.5 can handle difficult, long-running tasks that require creatively using tools to solve problems, whether in software engineering, data science, finance, legal work, or anything else you do on a computer.

This is a huge upgrade over Grok 4.3 👀

Meta launches Muse Image across its apps and previews Muse Video

ICYMI: Meta has rolled out Muse Image in Meta AI, bringing image generation, editing, and social creation to its apps. The model uses Instagram context, search, and coding tools, with wider rollout, ad access, and watermarking planned.

🗞 #meta @testingcatalog

ByteDance debuts Seedream 5.0 Pro with advanced reasoning

ByteDance Seed launched Seedream 5.0 Pro, a multimodal image model for production design. It supports dense layouts, region-specific edits, reusable assets, realistic visuals, and native text generation across 14 languages via partners and APIs.

🗞 #bytedance @testingcatalog

OpenAI rolls out GPT-Live voice for ChatGPT on web and mobile

OpenAI launched GPT-Live for ChatGPT Voice, rolling out globally from July 8, 2026. Its full-duplex design supports simultaneous listening and speaking, interruptions, and background task handling, with GPT-5.5 for harder requests.

🗞 #chatgpt @testingcatalog

SpaceXAI launches Grok 4.5 model on Grok Build and APIs

SpaceXAI launched Grok 4.5 for coding, agents, and knowledge work, now live in Grok Build, Cursor, and its API. It supports 500K context, tool use, code execution, and configurable reasoning, with EU access expected mid-July.

🗞 #grok @testingcatalog

Meta is working on “super sensing” AI glasses that can capture every moment, according to Financial Times.

> New glasses that would use cameras and audio recordings to capture a wearer’s every moment, as it pushes into the contentious market for all-seeing, all-hearing devices.

Meta doubles down on hardware 👀

Ultra-Long Video (Beta), powered by SeeDance 2.5 has been spotted on Jimeng.

This mode has been announced earlier and will allow users to generate 180 second long videos!

h/t x@MarsForTech

OpenAI set to launch ChatGPT Work upgrades today

OPENAI 🔥: ChatGPT Work will likely arrive in the form of an upgraded dedicated workspace, individual for every user.

What we know so far 👀

> "ChatGPT Work can help you build websites, prototype new ideas, create presentations and documents, and much more."

> Users will be able to “Reset” their workspaces without affecting their Library storage.

> There is a new Demo plugin that highlights features such as Email handling, Artifact creation, Learning writing style, Research, and Pet creation!

> Users will likely be able to create their own personalized Work Pets.

In general, ChatGPT Work looks like a new branding over Workspace solutions that will be powered by GPT-5.6 and a next step towards ChatGPT<>Codex consolidation.

🗞 #chatgpt @testingcatalog

Mistral unveils Robostral model for robot navigation with RGB camera

Mistral’s Robostral Navigate is an 8B embodied navigation model that uses one RGB camera and language instructions to guide robots. Trained in simulation, it leads R2R-CE with 76.6% success across varied robots and settings.

🗞 #mistral @testingcatalog

META 🔥: Muse Spark 1.1 is now listed on Meta Model API website.

> It comes with a big improvement over Muse Spark 1, with a 43-point difference on DeepSWE.

> However, the Meta Model API is not yet available in the EU.

Overall, Meta is one of the companies that can scale their AI to billions of users in a very short amount of time.

The fact that we saw a 1.1 upgrade coming so fast means only one thing: Meta is back 🤖

PERPLEXITY 🔥: A new GLM 5.2-powered preview orchestrator model is now available on Perplexity Computer.

The model delivers the same performance as Opus 4.8 at 0.344x the cost.

Tokensaving 👀

Meta debuts Muse Spark 1.1 model and opens API for developers

Meta launched Muse Spark 1.1 and the Meta Model API public preview, expanding access to its multimodal reasoning model with stronger agentic workflows, coding, computer use, long-context tasks, and added safety safeguards.

🗞 #meta @testingcatalog

Anthropic adds usage reflection dashboard to Claude for all users

Anthropic is rolling out a beta reflection dashboard for Claude that summarizes usage by topic, task, and timing, adds quiet hours and break prompts, and keeps reports private for Free, Pro, and Max users with Memory enabled.

🗞 #claude @testingcatalog