1487
Reporting AI nonsense. A future news media, driven by virtual assistants 🤖
xAI to add file attachment support in DeepSearch mode for Grok 3 on X
xAI plans to introduce file attachment support in DeepSearch mode for Grok 3 on X, allowing users to upload files like PDFs and images. This could help the AI incorporate user-provided data in web research. Grok 3 already supports attachments in its "Think" mode.
🗞 #grok
xAI prepares iOS app for voice mode release with new personalities
xAI is preparing to launch Grok Voice Mode on iOS, introducing a toggle for activation in settings. Updates include a notification for muted volume, personality selection instead of the transcribe menu, and a data export option for EU users. Availability remains limited.
🗞 #grok
Claude’s GitHub integration for Pro accounts moves closer to release
Anthropic is preparing updates for its cloud application, including a model announcement, reasoning and web search features, and LaTeX moving to general availability. GitHub integration may soon expand to personal accounts, allowing direct codebase access. A full rollout is expected around Amazon's event on February 26.
🗞 #claude
Grok Voice Mode enters internal testing as it gears up to rival ChatGPT and Gemini
xAI is preparing to launch Voice Mode for Grok, currently hidden in the iOS app but functional. The interface matches the Grok 3 announcement, with one available voice, Sal, and a dropdown suggesting future additions. Features include transcription, Internet toggle, and custom prompts. It supports singing and expressive speech but is expected to have limitations compared to ChatGPT’s Advanced Voice Mode. Powered directly by Grok 3, it is not a simple text-to-speech system. Elon Musk indicated a public rollout within one to two weeks, following internal testing.
🗞 #grok
Perplexity AI unlocks R1 1776 for developers with open-source release
Perplexity AI has open-sourced R1 1776, a post-trained version of DeepSeek-R1, addressing censorship and bias issues. The model, available on HuggingFace and via the Sonar API, was refined using Nvidia’s NeMo 2.0 to ensure factual, uncensored responses on sensitive topics.
🗞 #perplexity
Microsoft plans Copilot character feature launch in Japan
Microsoft is rolling out minor updates to Copilot, including character selection in Japan with Mica and Aqua as options. A side navigation bar for past chats and a share button for collaboration are also being tested. These changes refine usability and expand functionality.
🗞 #microsoftcopilot
World ID Orb verification launches in the Philippines to fight online fraud
World has launched its financial and identity network in the Philippines, introducing World ID to verify human identity online and counter AI-driven fraud. Piloted in Bulacan, it ensures privacy through advanced cryptographic methods. Filipinos can access services via the World App, gaining verification benefits while maintaining control over their data. This initiative addresses digital security challenges amid growing AI adoption and upcoming elections.
🗞 #worldcoin
Elon Musk confirms Grok 3 launch for Monday with live demo at 8 PM PT
Elon Musk confirmed the launch of Grok 3 on Monday at 8 PM PT, claiming it to be the smartest AI. Users speculate that two removed LM Arena models were test versions of Grok 3. It remains uncertain if Thinking capabilities will be available at launch.
🗞 #grok
Gemini can now summarize and link past chats with Recall feature
Google introduced "Recall" for Gemini, allowing users to reference past conversations and retrieve information quickly. A similar update for ChatGPT is expected next year. Additionally, Google is preparing updates for Gemini Live, promising a more natural conversational experience with vision capabilities.
🗞 #gemini
BREAKING 🚨: Perplexity is preparing a release of Deep Research Alpha 🔥
As this feature works on Gemini and ChatGPT, perplexity can take more time to go through web resources and utilise its reasoning capabilities to complete the report.
It seems like this feature has been temporarily available or it is only available to a small percentage of users. Considering its "Alpha" status it may happen that it would be released gradually.
In the video, I am asking to categorise Perplexity-related posts from TestingCatalog and it does quite a decent job. Speed is x4 👀
OpenAI developing search feature for Sora to navigate videos and images
OpenAI is adding a search feature to Sora, allowing users to find videos and images across a Media Library and an Explore section. The top feed is now functional but not public. Sora, launched in Dec 2024, creates videos from text and is available to ChatGPT Plus and Pro users.
🗞 #sora
Perplexity AI developing travel planning tools with hotel pages and AI travel agent
Perplexity AI is developing travel planning features, including Hotel Pages with Tripadvisor-sourced data and AI-generated summaries, and Travel Agent AI for personalized trip planning. These tools position Perplexity as a potential competitor to Google Travel and Expedia.
🗞 #perplexity
Google expands NotebookLM Plus to individual users with Google One AI Premium
Google has expanded NotebookLM Plus to individual users via the Google One AI Premium plan. It offers higher usage limits, advanced customization, and audio overviews for $19.99/month, with a 50% discount for students. The move reinforces Google's focus on AI-driven productivity tools.
🗞 #notebooklm
Mystery AI models on LM Arena: Could they be Grok 3 or Opus 3.5?
Two mysterious models, Chocolate and Kiwi, appeared on LM Arena, outperforming others while concealing their origins. Users speculate they could be Grok 3, o3, or Opus 3.5. Unusual SVG handling and structured image generation suggest advanced reasoning, but no confirmation exists.
🗞 #ai
Pikadditions by Pika Labs lets users seamlessly insert objects into videos
Pika Labs has introduced "Pikadditions" within Pika Turbo, allowing users to integrate objects or characters from images into videos naturally. This feature has been widely used for humorous content, and Pika Labs stands out as a tool for animating memes and cartoon characters.
🗞 #pika
Improved ChatGPT memory feature reaches more users on mobile and desktop
OpenAI is expanding the rollout of its improved memory feature, Moonshine, to randomly selected Plus and free users, now available on iOS and macOS. iOS users can also access ChatGPT without logging in. macOS updates may include UI changes and new tool tiles.
🗞 #chatgpt
Try Grok 3 for free as xAI gears up SuperGrok subscription
xAI has made Grok 3 temporarily free, allowing broader access before it transitions to a paid model. The company is also refining subscriptions, promoting X Premium Plus and introducing SuperGrok, a standalone option at $30/month. The platform's context capacity remains unclear.
🗞 #grok
xAI debuts Grok Personas from News to Unhinged as xAI blends utility with fun
xAI has introduced "Personas" for Grok, offering predefined roles like News, Doctor, Friend, Tutor, Romance, Unhinged, and Therapist. These roles cater to various user needs while raising ethical concerns, particularly for health-related personas, requiring careful oversight.
🗞 #grok
New Claude features signal major update ahead of Amazon’s February 26 event
Anthropic is preparing to release new reasoning features and web search for Claude, requiring an updated model. The iOS app now includes an Analysis tool, a new Tool Selector, and a beta Web Search toggle. A Pro upsell for “Extended Thinking” is also in development.
🗞 #claude
Grok 3 is now the most powerful AI model and you can try it for free
Elon Musk's xAI has launched Grok 3, a new AI model competing with GPT-4o and Gemini. It includes specialized reasoning variants and a DeepSearch feature. Trained on extensive computational resources, it's available via X’s Premium+ service with future API and voice mode updates.
🗞 #grok
Mistral Saba: New AI model specializes in Arabic and South Asian languages
Mistral AI has introduced Mistral Saba, a 24-billion-parameter regional language model tailored for the Middle East and South Asia. Supporting Arabic and Indian-origin languages, it prioritizes linguistic accuracy and cultural relevance. Available via API or local deployment, it offers efficient processing for conversational AI, domain-specific applications, and content creation.
🗞 #mistral
Claude’s next evolution? Anthropic may release an advanced reasoning model
Anthropic is preparing a reasoning model for Claude, referenced as Paprika Mode in code updates. A potential launch of Claude Opus 3.5 is expected soon, possibly aligning with Amazon’s February 26 event. Additionally, Anthropic is enhancing Google Drive search and testing new scrolling behavior.
🗞 #claude
OpenAI tests improved memory for ChatGPT as Google launches recall for Gemini
OpenAI is rolling out memory improvements under the “Moonshine” project, allowing ChatGPT to recall past chats. A business plan is in development, though details are scarce. GPT-4.5 is expected soon, while GPT-5 will integrate reasoning functions. o3's full release is postponed.
🗞 #chatgpt
Perplexity AI tests deep research feature, challenging Gemini and ChatGPT
Perplexity is launching Deep Research Alpha, allowing extended web analysis and reasoning for comprehensive reports. A recent test showed it categorizing posts accurately. An export feature for PDF and markdown has been added. Rollout details remain unclear.
🗞 #perplexity
Windsurf Wave 3: Codeium expands AI-assisted coding with MCP integration
Codeium has released Windsurf Wave 3, introducing MCP support, Turbo Mode, Tab-to-Jump, image drag-and-drop, and custom app icons. The update also includes new model integrations and early access to Windsurf Next, reinforcing Codeium’s focus on AI-driven developer tools.
🗞 #ai
Gemini now can access your past conversations. It can summarise previous chats and bring links to these chats as a response! 🔥
Читать полностью…
Ai2 launches open-source OLMoE iOS app for on-device AI models
Ai2 has launched the OLMoE iOS app, an open-source tool enabling on-device language model execution without internet reliance. Optimized for newer Apple devices, it balances efficiency and performance. The release supports research and development in local AI solutions.
🗞 #ai
Zyphra debuts Zonos, an open-source TTS suite with expressive voice cloning
Zyphra has released the beta version of Zonos-v0.1, an open-source TTS suite featuring high-fidelity voice cloning and real-time processing. It includes transformer and SSM-hybrid models under Apache 2.0, with broad multilingual support and advanced speech conditioning options. While offering competitive quality, Zonos faces challenges like audio artifacts and inference costs. Future updates aim to refine language support, pronunciation, and efficiency.
🗞 #ai
Windsurf wave 2: When AI actually gets how developers work
Windsurf’s Wave 2 update introduces a credit-based system and integrates DeepSeek R1 and OpenAI’s O3-mini models, delivering strong performance at lower costs. The Cascade system retains project context, improving workflow. Developers praise its codebase awareness, error handling, and VS Code integrations, with growing enterprise adoption.
🗞 #ai
I have never seen anything like this on lmarena before. Regardless of which model this is, it is worth testing (chocolate).
In my case, it glitched and started only responding with SVGs, as if it decided on its own that it was a better representation 👀
It was drawing them gradually 🤯