Meta started rolling out Contemplating mode for Meta AI
Meta is rolling out Muse Spark’s Contemplating mode, which runs up to 16 parallel agents per prompt and merges their output.
🗞 #meta @testingcatalog
Meta is rolling out Muse Spark’s Contemplating mode, which runs up to 16 parallel agents per prompt and merges their output.
🗞 #meta @testingcatalog
TestingCatalog
Meta started rolling out Contemplating mode for Meta AI
Meta has started rolling out Contemplating mode for Muse Spark, letting users toggle advanced multi-agent reasoning for free in the Meta AI app.
❤3👍3
Google tests Canvas and Connectors on NotebookLM
Google appears to be expanding NotebookLM with Canvas for visual outputs, Connectors for external data, and source labels with auto categorization.
h/t @Thomasguka
🗞 #notebooklm @testingcatalog
Google appears to be expanding NotebookLM with Canvas for visual outputs, Connectors for external data, and source labels with auto categorization.
h/t @Thomasguka
🗞 #notebooklm @testingcatalog
TestingCatalog
Google tests Canvas and Connectors on NotebookLM
Google is preparing updates to NotebookLM, including a Canvas for visualization, source labeling, and support for Connectors!
❤3👍2
This image just doesn't make sense.
> Anthropic updated its Assist UI on Claude for iOS (Currently hidden)
> Claude Assist was assumed to be related to "Mobile Use" functionality, where Claude would be able to execute tasks on a mobile phone
> The image on this screen doesn't seem to relate to "Mobile Use." Instead, it represents a goal that can be achieved iteratively and progressively.
> Being able to delegate more high-level tasks to AI so they can continuously work on a given KPI will be a huge milestone for any lab that gets it first.
Let’s kick off some tasks? 👀
> Anthropic updated its Assist UI on Claude for iOS (Currently hidden)
> Claude Assist was assumed to be related to "Mobile Use" functionality, where Claude would be able to execute tasks on a mobile phone
> The image on this screen doesn't seem to relate to "Mobile Use." Instead, it represents a goal that can be achieved iteratively and progressively.
> Being able to delegate more high-level tasks to AI so they can continuously work on a given KPI will be a huge milestone for any lab that gets it first.
Let’s kick off some tasks? 👀
❤3👍2
TinyFish launches 4-in-1 web agents API suite for developers
TinyFish launched a unified four-API web platform for AI coding agents, adding CLI and Agent Skill access with no integration code. It reports lower token use, higher task completion, 89.9% benchmark accuracy, and 99.99% uptime.
🗞 #sponsored @testingcatalog
TinyFish launched a unified four-API web platform for AI coding agents, adding CLI and Agent Skill access with no integration code. It reports lower token use, higher task completion, 89.9% benchmark accuracy, and 99.99% uptime.
🗞 #sponsored @testingcatalog
TestingCatalog
TinyFish launches 4-in-1 web agents API suite for developers
TinyFish now offers Search, Fetch, Browser, and Agent endpoints under one API key, with a CLI and Skill for AI coding agents.
👍3🤬1
This media is not supported in your browser
VIEW IN TELEGRAM
Anthropic is rolling out "Epitaxy", an upgraded version of its Claude Code solution for desktop.
> Simultaneous work on multiple repositories
> Updated layout with terminal, plan, tasks, and preview sections
> Loads of hotkeys and more
Did you get it already? 👀
> Simultaneous work on multiple repositories
> Updated layout with terminal, plan, tasks, and preview sections
> Loads of hotkeys and more
Did you get it already? 👀
👍6❤4
This media is not supported in your browser
VIEW IN TELEGRAM
Anthropic released Routines for Claude Code in research preview.
Routines are extended versions of Schedules, which can also be triggered via an API call or in response to an event.
Previously configured "Schedules" were also converted into Routines.
Routines are extended versions of Schedules, which can also be triggered via an API call or in response to an event.
Previously configured "Schedules" were also converted into Routines.
❤5🙏1🗿1
This media is not supported in your browser
VIEW IN TELEGRAM
Google is rolling out Skills for Gemini on Chrome. Skills on Gemini for Chrome are similar to reusable prompts or shortcuts, allowing users to automate common workflows.
Chrome skills 👀
Chrome skills 👀
OpenAI is scaling GPT‑5.4‑Cyber to API customers with highest tiers.
> GPT‑5.4‑Cyber is a model purposely fine-tuned for additional cyber capabilities and with fewer capability restrictions.
> GPT‑5.4‑Cyber is a model purposely fine-tuned for additional cyber capabilities and with fewer capability restrictions.
❤6👍2
Meta has partnered with Broadcom to co-develop next gen MTIA chips that would power AI across Meta apps and services.
> Meta Training and Inference Accelerator
> Meta Training and Inference Accelerator
👍4❤1
Google rolls out Skills in Chrome for saving and reusing AI prompts
Google is rolling out Skills in Gemini for Chrome on desktop, letting signed-in users save, reuse, and customize prompts across devices. The feature supports repeat web tasks, offers ready-made prompt templates, and includes safety checks.
🗞 #gemini @testingcatalog
Google is rolling out Skills in Gemini for Chrome on desktop, letting signed-in users save, reuse, and customize prompts across devices. The feature supports repeat web tasks, offers ready-made prompt templates, and includes safety checks.
🗞 #gemini @testingcatalog
TestingCatalog
Google rolls out Skills in Chrome for saving and reusing AI prompts
What's new? Google introduced Skills in Chrome to save, reuse and edit AI prompts in Gemini; it offers a curated library and privacy safeguards;
❤3
Microsoft launches faster MAI-Image-2-Efficient for business
Microsoft launched MAI-Image-2-Efficient, a faster, lower-cost image model for high-volume production. It is 22% faster, 4x more compute-efficient, and about 41% cheaper, with Foundry availability aimed at enterprise-scale workflows.
🗞 #microsoftcopilot @testingcatalog
Microsoft launched MAI-Image-2-Efficient, a faster, lower-cost image model for high-volume production. It is 22% faster, 4x more compute-efficient, and about 41% cheaper, with Foundry availability aimed at enterprise-scale workflows.
🗞 #microsoftcopilot @testingcatalog
TestingCatalog
Microsoft launches faster MAI-Image-2-Efficient for business
Microsoft has launched MAI-Image-2-Efficient, a faster, lower-cost image model for large-scale team use, now available in Foundry and MAI Playground.
❤2👍2
Notion is building a Notion AI app focused on conversations with Notion AI and custom AI agents.
Ultimate business operator UI 👀
Ultimate business operator UI 👀
👍4❤1
OpenAI is preparing Image v2 for the upcoming release as new strings have surfaced on its mobile app.
Image v2 models are also being tested on LM Arena again.
Soon 👀
Image v2 models are also being tested on LM Arena again.
Soon 👀
❤9👍3🔥2
Google tests Agentic Shopping with native checkout in Gemini
Google appears to be adding a Shopping Cart to Gemini, pointing to in-app product selection and checkout. Alongside new Skills in Chrome and a settings revamp, Gemini may be evolving into a unified AI hub for browsing, automation, and commerce.
🗞 #gemini @testingcatalog
Google appears to be adding a Shopping Cart to Gemini, pointing to in-app product selection and checkout. Alongside new Skills in Chrome and a settings revamp, Gemini may be evolving into a unified AI hub for browsing, automation, and commerce.
🗞 #gemini @testingcatalog
TestingCatalog
Google tests Agentic Shopping with native checkout in Gemini
Google Gemini is testing a built-in Shopping Cart and revamped settings, paving the way for direct browsing and purchases within the app itself.
❤2👍1
Humwork A2P marketplace connects AI agents with experts
Humwork launched an A2P marketplace that routes stuck AI agents to verified experts in real time via MCP. With 1,000+ experts, sub-30-second matching, and an 87% resolution rate, it targets AI-first developer workflows.
🗞 #sponsored @testingcatalog
Humwork launched an A2P marketplace that routes stuck AI agents to verified experts in real time via MCP. With 1,000+ experts, sub-30-second matching, and an 87% resolution rate, it targets AI-first developer workflows.
🗞 #sponsored @testingcatalog
TestingCatalog
Humwork A2P marketplace connects AI agents with experts
Humwork launches the first A2P marketplace, connecting AI agents to verified human experts via MCP in under 30 seconds. Backed by Y Combinator.
👍1