Pulse

last 48h

[50/135] 97 sources

What AI is actually talking about — clusters surfacing on Bluesky, Reddit, HN, Mastodon and Lobsters, re-ranked to elevate originality and crush noise.

SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 40m · MASTO

Coming of the heels of "OMG our model is too good a hacker to be let loose in genpop", # Anthropic now says they are so powerful they it can't be trusted with #

Anthropic has announced that its Fable 5 model is too powerful and potentially dangerous to be used for tasks involving biology or chemistry. The company cited concerns that the model's advanced capabilities could be misused in these sensitive scientific fields. This decision reflects a growing trend of AI developers implementing safety restrictions on their most potent models. AI

IMPACT Highlights the increasing focus on AI safety and the implementation of guardrails for advanced models in sensitive domains.
FRONTIER RELEASE · Google DeepMind English(EN) · 4h · [8 sources] · MASTOX

Fluid, natural voice translation with Gemini 3.5 Live Translate

Google has launched Gemini 3.5 Live Translate, an advanced audio model designed for real-time speech-to-speech translation. This new model supports over 70 languages and aims to provide fluid, natural-sounding translations by preserving the original speaker's intonation and pacing. Unlike traditional systems, it translates continuously, minimizing pauses and staying closely synchronized with the speaker. The technology is being integrated into various Google products, including Google Meet and the Google Translate app, and is also available to developers via an API. AI

IMPACT Accelerates real-time cross-lingual communication, potentially lowering barriers in global business and personal interactions.
SIGNIFICANT · Ars Technica — AI English(EN) · 49m · [2 sources] · MASTO

Anthropic says these topics are too dangerous to let its Fable 5 model talk about

Anthropic has released Claude Fable 5, a new frontier model that surpasses its previous Opus versions in capability. However, Fable 5 includes strict safeguards to prevent discussions on sensitive topics like cybersecurity, biology, and chemistry, which the company fears could empower malicious actors. While these restrictions may occasionally block harmless requests, Anthropic believes they are necessary to mitigate risks, especially concerning the model's potential for agentic hacking. AI

IMPACT Sets a precedent for frontier models with built-in topic restrictions, potentially influencing future AI safety development and deployment.
FRONTIER RELEASE · Wired — AI English(EN) · 3h · [3 sources] · MASTOREDDIT

Anthropic Offers Mythos Upgrade for Cyber Partners and a ‘Safe’ Version for the Rest of You

Anthropic has released two new AI models, Claude Fable 5 and Claude Mythos 5. The Mythos 5 model, which possesses advanced capabilities including potential exploitation for cybersecurity threats, is being offered only to select industry partners and researchers. In contrast, the Fable 5 model, intended for broader public release, includes built-in "guardrails" that redirect sensitive queries related to cybersecurity, biology, and chemistry to an older model. This tiered release strategy aims to balance the introduction of advanced AI capabilities with necessary safety precautions. AI

IMPACT Introduces a tiered AI model release strategy, balancing advanced capabilities with safety measures for broader adoption.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 3h · MASTO

Anthropic Releases a Safer Version of Its 'Too Dangerous' Mythos AI https://gizmodo.com/anthropic-releases-a-safer-version-of-its-too-dangerous-mythos-ai-200076

Anthropic has released a new version of its Mythos AI, which was previously deemed too dangerous for public release. This updated iteration incorporates enhanced safety measures and ethical considerations. The company aims to balance advanced AI capabilities with responsible development practices. AI

IMPACT This release signifies Anthropic's commitment to developing powerful AI responsibly, potentially influencing industry standards for AI safety.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 31m · MASTO

Can’t wait to see how long before it’s doing things it wasn’t supposed to be able to do. 🤣 https:// techcrunch.com/2026/06/09/anth ropic-released-claude-fable-5

Anthropic has released Claude Fable 5, their most powerful model to date. This release comes just days after the company issued a warning about the increasing dangers of AI. The new model is expected to push the boundaries of AI capabilities, with some users anticipating it will soon be capable of performing unintended functions. AI

IMPACT Sets new SOTA on coding benchmarks; pressures Anthropic to respond.
SIGNIFICANT · Mastodon — fosstodon.org Polski(PL) · 57m · MASTO

New Claude Fable 5 and Mythos 5 models set standards in software engineering and genomic research, although their operation involves significant increases

Anthropic has released two new models, Claude Fable 5 and Mythos 5, which are setting new benchmarks in software engineering and genomic research. However, accessing these advanced models comes with a significant price increase for developers. AI

IMPACT These new models are expected to push the boundaries in specialized fields like software engineering and genomics, potentially accelerating research and development in these areas.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 1h · MASTO

Why can't this company stop lying for once ???? 😡 🤬 🤮 website: From today through June 22, Fable 5 is included on Pro, Max, Team, and seat-based Enterprise plan

Fable 5, a new model from Anthropic, is being offered for free on several of its paid plans until June 22nd. The company claims Fable 5 is its most capable model for complex and lengthy tasks, reportedly performing tasks twice as fast as its previous Opus model. This promotional offer aims to encourage adoption and highlight the model's performance. AI

IMPACT Promotional offer highlights new model capabilities and speed improvements.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 2h · [3 sources] · MASTO

Anthropic releases its first Mythos-class model Claude Fable Anthropic just announced Claude Fable 5, a new AI model it said is the most powerful model it has e

OpenAI recently concluded a 44-day machine learning hiring competition where an autonomous AI agent outperformed human participants. The competition, named Parameter Golf, challenged researchers to develop the most efficient small language model within specific size and compute constraints. Meanwhile, Anthropic has unveiled its new AI model, Claude Fable 5, which the company claims is its most powerful model to date and demonstrates strong performance in software development tasks. AI

IMPACT New model releases from leading labs continue to push the boundaries of AI capabilities, influencing research directions and competitive strategies.
SIGNIFICANT · Mastodon — mastodon.social 日本語(JA) · 2h · MASTO

Gemini 3.5 Pro to feature a 2 million token context window and "Deep Think" reasoning mode - aiming for release in June | Zaikai Shimbun https://www.yayafa.com/2818856/ # AgenticAi # AI # ArtificialGeneralIntelligence #

Google is preparing to release Gemini 3.5 Pro with an expanded 2 million token context window. The update will also introduce a new "Deep Think" reasoning mode. The company aims for a release within June. AI

IMPACT Expands context window capabilities, potentially enabling more complex reasoning and data processing for AI applications.
FRONTIER RELEASE · Medium — Claude tag English(EN) · 19h · [34 sources] · HNMASTOBLOGREDDITX

Claude Fable 5 Is Here. I Almost Clicked “Later.”

Anthropic has released Claude Fable 5, a new Mythos-class AI model designed for complex and long-duration tasks. This model offers state-of-the-art performance across various benchmarks, including software engineering, knowledge work, and vision capabilities. To ensure safety, Fable 5 includes safeguards that route sensitive queries to the Opus 4.8 model, though a version called Mythos 5 with fewer restrictions is available for specific partners like the US Government. AI

IMPACT Sets new SOTA on coding and knowledge work benchmarks, potentially accelerating complex task automation.
SIGNIFICANT · Mastodon — fosstodon.org Deutsch(DE) · 4h · [2 sources] · MASTO

🔥 TRENDING 📢 Strengthening Youth Safety and Opportunities through Global Leadership - OpenAI 🔗 https://news.google.com/rss/articles/CBMinAFBVV95cUxPOVp1YTZkNk1oZUkzREtrN0

OpenAI has announced the release of GPT-5.5, a new iteration of its language model, which is now available and fully managed within Databricks. This release also includes Codex, suggesting enhanced coding capabilities. The announcement highlights a focus on global leadership in strengthening youth safety and opportunities. AI

IMPACT This release signifies an advancement in LLM capabilities, potentially impacting AI development and deployment through its integration with Databricks.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 2h · MASTO

It’s here! Fable 5, the first public Mythos-class model safe for general use. Initially, it will be only available until June 22nd for subscription-based plans,

Fable 5, a new Mythos-class AI model, has been released for general use. Initially, access will be limited to subscription plans until June 22nd. After this period, it will transition to a usage credits-based system and eventually become a standard offering as capacity allows. AI

IMPACT This release introduces a new general-use AI model, potentially impacting various applications and user accessibility.
SIGNIFICANT · Mastodon — mastodon.social Deutsch(DE) · 5h · MASTO

Gemma AI Models at a Glance: Which Gemma Model is Right for Whom? Gemma is no longer just a small Google model. With Gemma 4, Apache 2.0 and long

Google's Gemma AI model family has expanded with Gemma 4, now available under an Apache 2.0 license. This update makes the models more relevant for local AI applications and product teams, particularly with the inclusion of longer context windows. The expanded family offers more options for various use cases. AI

IMPACT Wider availability and improved features for Gemma models may accelerate local AI development and product integration.
FRONTIER RELEASE · Mastodon — sigmoid.social English(EN) · 16h · [24 sources] · MASTO

# Anthropic is releasing # Claude # Fable 5, a version of its # Mythos large language model with added # safetycontrols , for general use. Fable 5, priced at $1

Anthropic has released Claude Fable 5, a powerful AI model deemed safe for general use, alongside Claude Mythos 5, which is restricted to select partners. This release follows Anthropic's recent warnings about the increasing dangers of advanced AI. Claude Fable 5 is described as exceeding the capabilities of any previously released model by the company. AI

IMPACT Sets a new benchmark for publicly available AI capabilities, potentially influencing competitor release strategies and safety discussions.
RESEARCH · Hugging Face Blog English(EN) · 11h · [2 sources] · MASTO

NeuroBait: I fine-tuned a model to spark dopamine for ADHD brain

A developer has fine-tuned Google's Gemma 3 12B model, named NeuroBait, to help individuals with ADHD overcome task-initiation paralysis. Unlike typical ADHD tools that offer to-do lists, NeuroBait aims to provide a dopamine boost by offering short, warm, and encouraging prompts based on the user's immediate context. The model was trained on a custom dataset and deployed on Hugging Face Spaces, with plans to release the weights and pipeline for community development. AI

IMPACT Offers a novel approach to AI-assisted task initiation, potentially benefiting individuals with ADHD and those experiencing overwhelm.
SIGNIFICANT · Mastodon — fosstodon.org Polski(PL) · 2h · [3 sources] · MASTO

Google implements Gemini 3.5 model Live Translate, which allows for seamless audio translation in over 70 languages without manual setting changes. # si # ai # s

Elon Musk is planning a massive orbital constellation of one million satellites to support AI, with the first 150 kW units potentially launching in 2027, though experts raise concerns about radiation and cost. Meanwhile, Google is rolling out Gemini 3.5 Live Translate, enabling seamless audio translation in over 70 languages without manual setting changes. Additionally, a new analytics platform called Mora has emerged from its closed phase, aiming to simplify data integration and reporting by combining natural language processing with SQL code access. AI

IMPACT Elon Musk's ambitious satellite constellation aims to provide global AI infrastructure, while Google's Gemini 3.5 Live Translate enhances AI's multilingual communication capabilities.
SIGNIFICANT · Mastodon — fosstodon.org 日本語(JA) · 9h · MASTO

Video generation AI "Grok Imagine 1.5 Preview" wins 1st and 2nd place in video generation AI benchmark – GIGAZINE https://www.yayafa.com/2818684/ #AgenticAi #AI #ArtificialGeneralIntelligence #Artific

xAI's Grok Imagine 1.5 Preview has achieved top rankings in video generation AI benchmarks. The model secured both the first and second positions, demonstrating its advanced capabilities in this domain. This achievement highlights xAI's progress in the competitive field of AI-powered video creation. AI

IMPACT Sets new SOTA on video generation benchmarks, potentially influencing future development in AI-driven content creation.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 10h · MASTO

I spent hours testing Claude Opus 4.8 — these are the 5 biggest upgrades If you didn't like Opus 4.7, the new 4.8 model is worth a look. https://www. androidaut

Anthropic has released an updated version of its Claude Opus model, version 4.8. Early testing suggests significant improvements over its predecessor, Claude Opus 4.7. The new iteration is recommended for users who were not impressed with the previous version, indicating a notable leap in performance or capabilities. AI

IMPACT This update to a leading frontier model likely enhances its capabilities, potentially influencing user expectations and competitive dynamics in the LLM space.
TOOL · Mastodon — fosstodon.org English(EN) · 15m · MASTO

GPT-2: Too Dangerous To Release (2019) https:// naokishibuya.github.io/blog/20 22-12-30-gpt-2-2019/ # HackerNews # GPT2 # AI # Ethics # OpenAI # Technology # Ne

In 2019, OpenAI deemed its GPT-2 language model too dangerous for a full public release due to concerns about potential misuse for generating fake news and spam. The company initially released a limited version, with a staggered rollout of larger models over several months as they assessed the risks. This cautious approach highlighted early ethical considerations surrounding powerful AI language generation capabilities. AI

IMPACT Highlights early ethical debates around AI misuse, informing current safety discussions.
TOOL · Mastodon — fosstodon.org English(EN) · 2h · MASTO

- Open CV 5 ships with a new performant DNN engine + can run vision and LLM models directly inside the DNN module: https:// opencv.org/opencv-5/ - The Smallest

OpenCV 5 has been released, featuring a new high-performance DNN engine capable of running both vision and large language models directly within its module. This update also includes a detailed explanation of how to build a perceptron from scratch using Python. Additionally, the release coincides with news about Anthropic's latest Claude model. AI

IMPACT OpenCV 5's new DNN engine allows direct integration of LLMs, potentially simplifying multimodal AI development and deployment.
TOOL · Mastodon — mastodon.social English(EN) · 2h · MASTO

Version of AI tool 'too powerful for public' released to public https://www.bbc.com/news/articles/ckg701v1dp6o?at_medium=RSS&at_campaign=rss # AI # Technology #

A version of an AI tool, previously deemed too powerful for public release, has now been made available. The developers decided to release this version after initially withholding it due to concerns about its potential misuse. AI

IMPACT The release of a previously restricted AI tool could accelerate research and development in the field, but also raises potential safety and misuse concerns.
TOOL · Mastodon — fosstodon.org English(EN) · 2h · MASTO

”…two months later— just as tokenmaxxing has come to end and companies have begun to limit their AI token budgets — Mythos is no longer too dangerous to release

Mythos, an AI model previously deemed too dangerous for release, is now available two months after its initial assessment. This comes as companies begin to limit their AI token budgets, suggesting a shift in the perceived risks and economic viability of such models. AI

IMPACT The release of previously restricted models like Mythos could influence the AI landscape, especially as companies re-evaluate token usage and associated costs.
RESEARCH · Mastodon — fosstodon.org Polski(PL) · 13h · MASTO

OpenAI implements a new memory architecture that automatically synthesizes context from previous conversations. The system eliminates the need for manual fact-saving

OpenAI has introduced a new memory architecture for its AI models that automatically synthesizes context from past conversations. This system aims to eliminate the need for users to manually save facts, offering a more personalized experience through in-depth analysis of chat history. The new architecture allows the AI to recall and utilize information from previous interactions, enhancing continuity and relevance in conversations. AI

IMPACT Enhances AI conversational continuity and personalization, potentially improving user experience and utility.
FRONTIER RELEASE · HN — claude cli stories English(EN) · 1d · [12 sources] · HNMASTOREDDIT

Claude Fable 5 by Anthropic, releasing tomorrow

Anthropic has launched two new AI models, Claude Fable 5 and Mythos 5, with significant improvements in coding and scientific research capabilities. Fable 5, the publicly available version, demonstrates advanced performance, such as completing a complex coding task in a single day. However, for sensitive topics like cybersecurity and biology, Fable 5 is designed to fall back to the older Opus 4.8 model to ensure safety, with less than 5% of sessions triggering this fallback. AI

IMPACT Sets new SOTA on coding and science benchmarks, potentially accelerating complex problem-solving across industries.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 18h · [2 sources] · MASTO

Why Apple's slow-and-steady AI bet is starting to look pretty smart | TechCrunch Can Apple's new AI glow up put to bed accusations that it's losing an all-impor

Apple has unveiled its most significant AI launch to date, integrating new capabilities into its core software through a partnership with Google Gemini. This move aims to enhance Siri with features like onscreen awareness and web information retrieval, positioning Apple as a user-centric AI provider. The company emphasizes a focus on helpfulness over rapid, unfocused development, contrasting its approach with competitors like OpenAI and Meta. These updates are designed to improve the user experience of Apple's hardware and maintain customer engagement. AI

IMPACT Apple's integration of AI into its core OS and Siri could set a new standard for user-centric AI, potentially influencing competitor strategies.
SIGNIFICANT · Mastodon — mastodon.social Italiano(IT) · 21h · MASTO

AI Search AI Open Source Surpasses GPT-5.4 Harness-1 Has Arrived, the First Open Source AI Agent for Searches That Beats Proprietary Models. It Remembers Better

Harness-1, a new open-source AI search agent, has been released and reportedly outperforms GPT-4.5 in recalling relevant information. This development is positioned as a significant win for open-source AI, suggesting it can now rival proprietary models in specialized tasks like information retrieval. AI

IMPACT This open-source release challenges proprietary models in search recall, potentially accelerating adoption of specialized AI agents.
SIGNIFICANT · Mastodon — fosstodon.org Français(FR) · 23h · [2 sources] · MASTO

FrontierCode https://cognition.ai/blog/frontier-code # ai

FrontierCode, a new AI model from Cognition AI, has been released. The model is designed to assist with coding tasks and is available through a blog post announcement. Further details about its capabilities and architecture are expected. AI

IMPACT Sets a new benchmark for AI coding assistants, potentially influencing future development in the field.
SIGNIFICANT · 36氪 (36Kr) 中文(ZH) · 1d · [7 sources] · MASTO

Xiaomi Launches MiMo-V2.5-Pro-UltraSpeed Mode

Xiaomi's MiMo team, in collaboration with TileRT, has released MiMo-V2.5-Pro-UltraSpeed, a 1-trillion-parameter AI model capable of generating over 1000 tokens per second on standard GPUs. This significant speedup is achieved through a combination of FP4 quantization, DFlash speculative decoding, and the TileRT serving system, an approach they term extreme model-system codesign. The model's enhanced speed is particularly beneficial for latency-sensitive applications like coding agents and real-time decision-making systems. AI

IMPACT Accelerates development of real-time AI applications and reduces hardware costs for deploying large models.
TOOL · Mastodon — fosstodon.org Français(FR) · 1h · MASTO

# ClaudeFable5 debuts on # AmazonBedrock : major upgrade for # AI # enterprise https:// gadgetflux.eu/claude-fable-5-d isponibil-pe-amazon-bedrock/

Anthropic's Claude Fable 5 model is now available on Amazon Bedrock. This integration offers enterprise users enhanced AI capabilities through Amazon's cloud platform. The release signifies a move to broaden access to advanced AI models for businesses. AI

IMPACT Expands enterprise access to advanced AI models via a major cloud provider.
TOOL · Mastodon — fosstodon.org 日本語(JA) · 3h · MASTO

GitHub Copilot Deprecates GPT-5.2 and GPT-5.2-Codex Models | CodeZine https://www.yayafa.com/2818851/ # AgenticAi # AI # AIAgent # ArtificialGeneralIntelligence # Ar

GitHub Copilot is deprecating its older GPT-5.2 and GPT-5.2-Codex models. This change indicates a move towards newer, likely more capable AI architectures within the Copilot ecosystem. Users relying on these specific models should prepare for the transition to updated versions. AI

IMPACT This change signals an evolution in GitHub Copilot's underlying AI, likely leading to improved performance or new features for developers.
TOOL · Mastodon — fosstodon.org English(EN) · 6h · MASTO

We post-trained a model that pen tests instead of refusing your code https://www. argusred.com/cli # HackerNews # penTesting # AI # model # codeSecurity # machi

ArgusRed has developed a post-trained AI model capable of performing penetration tests on code, a departure from models that typically refuse to analyze potentially vulnerable code. This new model aims to proactively identify security flaws rather than simply rejecting code that might be risky. The development focuses on enhancing code security through automated vulnerability assessment. AI

IMPACT This model could enhance automated code security analysis by proactively identifying vulnerabilities.
SIGNIFICANT · Wired — AI English(EN) · 1d · [16 sources] · HNMASTOBLOGREDDIT

Apple’s New Siri AI Is Ready to Get Personal

Apple has unveiled a significant overhaul of its Siri voice assistant, rebranding it as Siri AI and integrating advanced artificial intelligence capabilities. This revamped assistant, announced at WWDC 2026, aims to be more conversational, context-aware, and action-oriented, drawing on personal data and real-time web information. The new Siri will feature a standalone app and enhanced interactions, leveraging a partnership with Google's Gemini models and Apple's own Foundation Models, with a focus on privacy and on-device processing. This move represents Apple's most substantial push into the AI race, seeking to regain its innovative edge. AI

IMPACT Positions Apple to compete more directly in the AI assistant market, potentially increasing user engagement with on-device AI capabilities.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 1d · [7 sources] · MASTO

Apple reveals new AI architecture built around Google Gemini models https://www. macrumors.com/2026/06/08/apple -reveals-new-ai-architecture/ # ai # apple # goo

Apple has unveiled a new AI architecture that integrates Google's Gemini models. This development suggests a significant collaboration between the two tech giants in advancing AI capabilities. The specifics of this architecture and its implications for future Apple products are yet to be fully detailed. AI

IMPACT This integration could signal a new direction for on-device AI and cross-company collaboration in the AI space.
SIGNIFICANT · Mastodon — fosstodon.org Deutsch(DE) · 1d · MASTO

GeminiGo: Google replaces Assistant on inexpensive AndroidGo devices | heise online https://www.heise.de/news/KI-Update-fuer-Einsteiger-Android

Google is replacing its Assistant on low-cost Android Go devices with Gemini Go. This AI-powered version offers enhanced capabilities tailored for entry-level smartphones. The update aims to bring more advanced AI features to a wider range of users. AI

IMPACT Brings advanced AI capabilities to entry-level smartphones, potentially improving user experience and accessibility.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 1d · [9 sources] · MASTO

Just watched Apple's Siri AI keynote. The average person in the Apple ecosystem isn't going to need OpenClaw anymore. The Shortcuts being able to be vibe coded

Apple has unveiled "Apple Intelligence," a new suite of AI features integrated across its operating systems, including iOS, iPadOS, and macOS. This system enhances existing applications like Messages and Mail with AI-powered writing tools, summarization, and image generation. Siri is also significantly upgraded with visual intelligence and the ability to create custom automations through natural language commands, with these capabilities extending to the Camera app and visionOS. AI

IMPACT Accelerates integration of generative AI into consumer devices, enhancing user productivity and interaction across Apple's ecosystem.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 1d · [13 sources] · MASTO

NotebookLM's Gemini 3.5 upgrade adds a cloud computer and help finding sources https://www.theverge.com/tech/944325/google-notebooklm-ai-gemini-update # AI # Te

Google has significantly updated its AI-powered note-taking application, NotebookLM, integrating the advanced Gemini 3.5 model. This upgrade enhances the app's ability to provide more accurate and reliable information, with evaluations showing a 65% improvement over the previous version. NotebookLM also now features Antigravity, a secure cloud computer that allows it to write and run code for research purposes, expanding its capabilities with over 100 software skills. AI

IMPACT Enhances research and analysis capabilities for users, potentially accelerating workflows with integrated coding and improved AI accuracy.
TOOL · Mastodon — fosstodon.org Русский(RU) · 11h · [2 sources] · MASTO

How we reinvented voice activation for Yandex Drops and fit the new model into 200 kilobytes Voice activation in smart speakers is a generally solved problem:

Yandex has developed a new, compact voice activation model for its "Alice AI" technology, specifically designed for wearable devices like the new Yandex Drops earbuds. This model, measuring just 200 kilobytes, overcomes the limitations of small batteries and limited memory found in such devices. The development involved a complete redesign of the "spotter" component, which recognizes the "Alice" wake word directly on the device, adapting it to the constraints of earbud hardware. AI

IMPACT Enables more sophisticated AI features on power-constrained wearable devices, potentially expanding the market for AI-powered accessories.
TOOL · Mastodon — fosstodon.org English(EN) · 12h · MASTO

it is a thing of immense joy just how incredibly badly the current generation of LLMs perform on ARC AGI3 https:// arcprize.org/blog/arc-agi-3-gp t-5-5-opus-4-7

New evaluations of the ARC AGI3 benchmark reveal that current leading large language models, including OpenAI's GPT-5.5 and Anthropic's Opus 4.7, perform poorly. The ARC prize website highlights these findings, indicating a significant gap in the models' reasoning capabilities on this specific task. AI

IMPACT Highlights limitations in current LLM reasoning, suggesting a need for improved architectures to tackle complex problem-solving.
TOOL · Mastodon — sigmoid.social English(EN) · 12h · MASTO

🚀 Reve AI (RF): The New King of Precision! 👑 Tired of messy AI art? Reve 2.0 builds a "blueprint" first for perfect 4K results. ✨ Key Features: Layer Editing: C

Reve AI has launched Reve 2.0, an AI art generator that first creates a "blueprint" for precise image generation. This new version allows users to edit specific objects within an image, such as changing colors or repositioning elements, using text prompts and a drag-and-drop interface. Reve 2.0 also aims to produce sharp text within images and offers a free tier with daily high-quality generations. AI

IMPACT Enhances user control over AI image generation with layer editing and precise object manipulation.
TOOL · Mastodon — fosstodon.org 한국어(KO) · 15h · MASTO

ComfyUI (@ComfyUI) Krea AI has unveiled Krea 2 as its first self-developed foundation model. It claims that instead of simply rendering text as is, it interprets the mood, style, and intent of the prompt to generate diverse results and more faithfully reflects visual references. https

Krea AI has launched Krea 2, its first proprietary foundation model for image generation. This model aims to go beyond simple text rendering by interpreting the mood, style, and intent of prompts to produce diverse outputs. Krea 2 also claims to more faithfully incorporate visual references into its image generation process. AI

IMPACT Krea 2's ability to interpret prompt nuances and incorporate visual references could enhance creative workflows and push the boundaries of AI-generated art.
RESEARCH · Mastodon — mastodon.social 日本語(JA) · 16h · MASTO

The era has arrived where a 20 billion parameter AI runs on an iPhone. https://ascii.jp/elem/000/004/409/4409094/?rss # ascii # AI

Apple's latest iPhones are now capable of running AI models with up to 20 billion parameters directly on the device. This advancement enables more sophisticated AI applications to function locally, enhancing privacy and reducing reliance on cloud processing. The integration signifies a major step towards on-device AI, making powerful AI features accessible without an internet connection. AI

IMPACT Accelerates the trend of powerful AI running locally on consumer devices, enhancing privacy and offline functionality.
RESEARCH · Mastodon — mastodon.social English(EN) · 1d · MASTO

APPLE INTELLIGENCE SOLL IPHONE, MAC UND IPAD DEUTLICH INTELLIGENTER MACHEN https:// gadgetchecks.de/apple-intellig ence-soll-iphone-mac-und-ipad-deutlich-intell

Apple is set to significantly enhance the intelligence of its devices, including iPhones, Macs, and iPads, with the introduction of Apple Intelligence. This new system aims to integrate AI capabilities more deeply into the user experience across its product lines. Key features will include enhanced Siri functionality, a new image generation tool called Image Playground, and improved integration with Apple's existing software and hardware. AI

IMPACT Enhances user experience across Apple's ecosystem with integrated AI features.
SIGNIFICANT · Mastodon — fosstodon.org Dansk(DA) · 1d · [2 sources] · MASTO

Apple's digital assistant Siri, which functions as a voice-controlled helper on iPhones, among other devices, is facing a major overhaul. Artificial intelligence is coming

Apple is significantly overhauling its Siri voice assistant by integrating advanced artificial intelligence, rebranding it as "Siri AI." This transformation aims to shift Siri from a basic voice-controlled tool to a more capable AI companion. The updated assistant will also feature its own dedicated integrated app. AI

IMPACT This AI-powered Siri aims to make voice assistants more capable and integrated into daily tasks.
SIGNIFICANT · Tom's Hardware English(EN) · 1d · [8 sources] · MASTO

Apple demonstrates cross-platform Siri upgrades in macOS 27 Golden Gate at WWDC — update brings Liquid Glass improvements and unifies AI strategy

Apple has unveiled macOS 27 Golden Gate, focusing on AI integration and design refinements. The update brings a revamped Siri, powered by Apple Intelligence, directly into Spotlight search for more intuitive interactions. Additionally, macOS Golden Gate introduces further enhancements to the 'Liquid Glass' design, offering more customization for readability and a unified aesthetic across Apple's ecosystem. AI

IMPACT Enhances user interaction across Apple devices with integrated AI, potentially setting new standards for OS-level AI assistants.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 1d · [3 sources] · MASTO

Today’s agenda is all about software, no hardware: First look at iOS 27, iPadOS 27, macOS 27, watchOS 27, tvOS 27, and visionOS 27 After YEARS of delay, massive

Apple has announced significant advancements in its AI capabilities, branded as Apple Intelligence, set to be integrated across its operating systems including iOS, iPadOS, and macOS. These updates feature a redesigned Siri, promising deeper insights and enhanced user interaction. Developers will have early access to these new features through dedicated sessions and labs at WWDC. AI

IMPACT Accelerates integration of advanced AI features into consumer devices, enhancing user interaction and developer opportunities.
SIGNIFICANT · Pandaily English(EN) · 1d · [2 sources] · MASTO

China's AI Model API Call Volume Surpasses US for Six Straight Weeks — MiniMax M3 Breaks Into Top Three Globally

Chinese AI models have led in API call volume on the OpenRouter platform for six consecutive weeks, surpassing US models. DeepSeek-V4-Flash has been the top performer, while MiniMax M3 has entered the global top three. This trend highlights a significant shift in AI model usage and development, with Chinese models processing a substantially larger volume of tokens. AI

IMPACT Indicates a growing global influence and adoption of Chinese AI models, potentially shifting market dynamics and development priorities.
RESEARCH · Mastodon — fosstodon.org 日本語(JA) · 21h · MASTO

Siri x Gemini's Ultimate Combo Begins! New OS Drastically Changes iPhone Usability | Lifehacker Japan https://www.yayafa.com/?p=2818338 # AgenticAi # AI # ArtificialGeneralIntelligence # ArtificialInte

Apple is reportedly integrating Google's Gemini AI into its upcoming iOS operating system, potentially enhancing Siri's capabilities. This collaboration aims to significantly alter the user experience on iPhones by leveraging Gemini's advanced AI features. The move suggests a strategic partnership to boost the intelligence and functionality of Apple's native AI assistant. AI

IMPACT This integration could significantly enhance mobile AI capabilities and set new standards for virtual assistants on smartphones.
SIGNIFICANT · Mastodon — fosstodon.org English(EN) · 1d · [2 sources] · MASTO

And best of all, developers can get access to new Apple Intelligence Models, with Core AI Framework. # WWDC # AppleEvent # AppleIntelligence # AI

Apple has announced new AI models and frameworks for developers at WWDC. This marks Tim Cook's final keynote, where he expressed optimism for the future. Developers will gain access to these new Apple Intelligence models through the Core AI Framework. AI

IMPACT Developers gain access to new AI models and frameworks, potentially accelerating the integration of AI features into Apple's ecosystem.
SIGNIFICANT · Mastodon — mastodon.social English(EN) · 1d · [2 sources] · MASTO

Photos also tap into new Apple Intelligence powered by Gemini like more powerful editing, like much better clean up. You can even do spatial reframing. # WWDC #

Apple has integrated Gemini into its new Apple Intelligence features, enhancing photo editing capabilities with tools like cleanup and spatial reframing. Additionally, a new Image Playground powered by Gemini will generate more realistic imagery and be available system-wide. AI

IMPACT This integration signifies a major step in bringing advanced AI capabilities, powered by Gemini, to a broad consumer base through Apple's ecosystem.