PulseAugur / Pulse
LIVE 09:11:43

Pulse

last 48h
[50/118] 89 sources

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

  1. Nous Research Releases Token Superposition Training to Speed Up LLM Pre-Training by Up to 2.5x Across 270M to 10B Parameter Models

    Nous Research has developed Token Superposition Training (TST), a new method designed to significantly accelerate the pre-training of large language models. This technique can reduce pre-training time by up to 2.5 times for models ranging from 270 million to 10 billion parameters, without altering the model's architecture or how it performs inference. TST achieves this by modifying the training loop in two phases: an initial 'superposition' phase where token embeddings are averaged and processed in larger bags, followed by a 'recovery' phase that reverts to standard training. Experiments showed TST achieving lower final training loss with substantially less compute time compared to traditional methods. AI

    Nous Research Releases Token Superposition Training to Speed Up LLM Pre-Training by Up to 2.5x Across 270M to 10B Parameter Models

    IMPACT Accelerates LLM pre-training, potentially reducing compute costs and time for developing new large language models.

  2. @matthewberman on YT! Everyone's getting hacked # AI # Cybersecurity # Mythos 5/13/2036 https:// youtu.be/hAzhVloGkOw?si=03S2wO Es3_iflQzp

    The UK's AI Security Institute has released findings on new AI models, noting significant gains in cyber capabilities from both Mythos and GPT-5.5. These models appear to be limited by token usage rather than inherent ability, with a capability doubling time estimated at 4.5 months. Separately, Palantir CEO Alex Karp criticized Germany's defense procurement, urging them to adopt battle-tested Ukrainian technology. AI

    IMPACT New AI models show rapid capability doubling, potentially impacting cybersecurity and defense technology procurement.

  3. 🤖 AI Weather Forecasts Challenge Traditional Methods Forget everything you know about weather apps. New AI models like Graphcast, Aurora, and Pangu Weather are

    AI models such as Graphcast, Aurora, and Pangu Weather are emerging as alternatives to traditional weather forecasting methods. These new systems aim to provide faster and potentially more accurate predictions than conventional approaches. Their development signifies a shift towards leveraging advanced AI for complex environmental modeling. AI

    🤖 AI Weather Forecasts Challenge Traditional Methods Forget everything you know about weather apps. New AI models like Graphcast, Aurora, and Pangu Weather are

    IMPACT AI models are beginning to offer competitive alternatives to established methods in complex domains like weather prediction.

  4. "The developers I talked to agreed that LLMs will stick around and play a role in programming in the future in some fashion, but worried about how the industry

    Frontier AI models are showing a rapid increase in their ability to handle complex tasks, with their reliability doubling every 4.7 months, a rate that has accelerated since late 2024. Recent models like Claude Mythos Preview and GPT-5.5 are outperforming these trends, though their exact capabilities are still being measured due to near-perfect success rates on current benchmarks. This rapid progress challenges existing testing methodologies, as models are pushing the limits of token capacity and agent scaffolding, making it difficult to accurately assess their performance and potential deterioration at scale. AI

    IMPACT Rapid advancements in frontier models may necessitate new evaluation methods and could accelerate the adoption of AI in complex domains.

  5. 📰 Uncensored AI Model SuperGemma 26B: Local Usage Guide 2026 SuperGemma 26B is an AI model that stands out with its completely uncensored structure. Ollama

    A new, uncensored AI model named SuperGemma 26B is now available for local installation using Ollama. Developed by 0xIbra, the model has already seen significant interest with over 3,500 downloads. Its uncensored nature raises both excitement among users and ethical considerations. AI

    📰 Uncensored AI Model SuperGemma 26B: Local Usage Guide 2026 SuperGemma 26B is an AI model that stands out with its completely uncensored structure. Ollama

    IMPACT Provides a new, uncensored model for local experimentation, potentially enabling novel applications but also raising ethical concerns.

  6. Meta's Muse Spark won't be open-sourced, citing safety concerns over chemical and biological capabilities. This marks a shift: Meta now treats openness as a dep

    Meta has decided not to open-source its Muse Spark AI model, citing safety concerns related to its potential for misuse in chemical and biological applications. This decision represents a strategic shift for Meta, moving away from a principle of open-sourcing towards a more selective approach based on deployment safety. The model is slated for integration into Meta's own platforms and devices, such as its augmented reality glasses. AI

    IMPACT Meta's decision to keep Muse Spark closed signals a growing trend of frontier AI labs prioritizing safety over open access, potentially impacting the broader AI research community.

  7. https://nextlogic-ai.achlabo.com/en/%e3%80%8c%e7%be%8e%e3%80%8d%e3%81%ae%e8%a3%8f%e5%81%b4%ef%bc%9a%e3%82%b3%e3%83%a9%e3%83%bc%e3%82%b2%e3%83%b3%e3%82%b5%e3%8

    NextLogic AI has released a new model that can generate color images from text prompts. This model is designed to assist in various fields, including biotechnology and nutrition, by providing visual representations of concepts. The company aims to leverage AI for creative and scientific applications. AI

    IMPACT Enables visual representation of scientific concepts, potentially accelerating research and development in fields like biotech and nutrition.

  8. Unity AI Open Beta Unity has released Unity AI Beta, an AI tool specialized for game development. Unity AI consists of in-editor agents, AI Gateway, and MCP servers, providing optimized support within the Unity project context, and is available in Unity 6.0 and above.

    Unity has launched a public beta for its suite of AI tools designed specifically for game development. These tools, including an in-editor agent, AI Gateway, and MCP server, are optimized for Unity projects and require Unity 6.0 or later. The service offers a free tier for personal users with a paid subscription option, and is included for Pro and Enterprise subscribers, also allowing integration with third-party AI models. AI

    IMPACT Provides game developers with specialized AI tools to streamline asset creation and project workflows.

  9. Wes Roth (@WesRoth) refutes Andrew Ng's 'jobpocalypse' narrative that AI will cause mass unemployment soon, emphasizing that AI will transform work methods and roles rather than replace jobs. The message is that realistic transition and adaptation are needed instead of excessive fear. https:/

    Microsoft Research has unveiled GridSFM, a compact foundation model designed to optimize power grid efficiency. This model can predict optimal AC power flow in milliseconds, aiding operators in managing grid congestion, stability, and overall system health for cost savings. Separately, Andrew Ng refutes the notion of an imminent "jobpocalypse" due to AI, asserting that AI will transform rather than replace jobs, necessitating adaptation over excessive fear. AI

    IMPACT GridSFM's predictive capabilities could enhance power grid efficiency and cost savings, while Andrew Ng's commentary addresses the evolving nature of work in the age of AI.

  10. StepFun (@StepFun_ai) Step Image Edit 2 has been released, with a new version of the image editing model now available in real-time. This 3.5B parameter image model ranked first in all categories (overall, faithfulness, and concept) on the KRIS-Bench, an instruction-based image editing benchmark.

    StepFun has released Step Image Edit 2, a 3.5 billion parameter image editing model that has achieved top rankings on the KRIS-Bench benchmark across multiple categories. This new version surpasses significantly larger models in performance and offers a rapid response time of 0.7 seconds. Concurrently, Tencent's Hy AI model is now available in preview on gmi_cloud, allowing developers to test its latest features. AI

    IMPACT New image editing and generative models are released, with Step Image Edit 2 setting new benchmarks and Tencent offering early access to its Hy3 model for developer testing.

  11. ⚙️ New Ollama Release! ⚙️ Version: v0.23.4 Release Notes: ## What's Changed * `ollama launch opencode` now supports vision models with image inputs * Fixed form

    Ollama has released version 0.23.4, introducing support for vision models with image inputs when launching the opencode model. This update also addresses an issue with the formatting of Claude tool results when local image paths are used. AI

    IMPACT Enables broader local use of multimodal AI models through improved tooling.

  12. Apple's drawn pie, Google gets it done first! Gemini fully enters the whole family bucket, even the mouse is AI-powered.

    Google has integrated its Gemini AI into the Android operating system, enabling system-level services across applications and devices. This new Gemini Intelligence feature allows for contextual understanding and task execution, aiming to replicate the cross-device AI experience previously anticipated from Apple. The company also unveiled a "Magic Pointer" mouse cursor that uses AI for semantic understanding and task execution via voice commands and gestures, alongside a new "Googlebook" laptop designed to showcase these Gemini capabilities. AI

    IMPACT Accelerates AI integration into everyday computing, potentially setting a new standard for cross-device AI experiences.

  13. When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction

    A new research paper introduces a "channel-transition" framework to explain why large language models struggle to maintain context and instructions over extended multi-turn conversations. The study proposes the Goal Accessibility Ratio (GAR) as a metric to quantify the degradation of attention to key instructions. Researchers found that while attention to instructions may close, relevant information can persist in residual representations, leading to varied failure modes across different model architectures. AI

    IMPACT Identifies a core limitation in LLM conversational ability, potentially guiding future architectural improvements for better long-term memory.

  14. Meet AntAngelMed: A 103B-Parameter Open-Source Medical Language Model Built on a 1/32 Activation-Ratio MoE Architecture

    Researchers have introduced AntAngelMed, a 103 billion parameter open-source medical language model. It utilizes a Mixture-of-Experts (MoE) architecture, activating only 6.1 billion parameters per query for enhanced efficiency. This design allows it to match the performance of a 40 billion parameter dense model while achieving speeds over 200 tokens per second on H20 hardware. The model supports a 128K context length and has undergone a three-stage training process including pre-training on medical corpora, supervised fine-tuning, and reinforcement learning. AI

    Meet AntAngelMed: A 103B-Parameter Open-Source Medical Language Model Built on a 1/32 Activation-Ratio MoE Architecture

    IMPACT Provides a highly efficient, open-source LLM for medical applications, potentially accelerating research and development in the healthcare sector.

  15. SubQ is a new "subquadratic" LLM that can handle context windows of 12 million tokens. 12 million tokens is a massive amount of text, roughly equivalent to 9 mi

    A new large language model named SubQ has been announced, boasting the ability to process context windows of up to 12 million tokens. This represents a significant leap in context handling, potentially equivalent to hundreds of novels. The model also claims to offer 52 times faster AI inference speeds, though details on its cost and performance are still emerging. AI

    IMPACT Potentially enables new classes of applications requiring deep understanding of long documents or conversations.

  16. Let's Verify Step by Step compares process and outcome supervision on MATH. The process-reward model reaches 78.2% best-of-1860 vs 72.4% for outcome. But that g

    Researchers have developed SCoRe, a novel two-stage reinforcement learning technique that enables language models to refine their own responses using self-generated data. This method significantly improves performance on benchmarks like MATH and HumanEval when applied to models such as Gemini 1.5 Flash and 1.0 Pro. Additionally, a separate study explored process versus outcome supervision for mathematical reasoning, finding that process-reward models yield better results, though the advantage diminishes with fewer samples. AI

    IMPACT New self-correction techniques could enhance LLM reasoning capabilities and reduce the need for extensive human supervision in training.

  17. Show HN: Statewright – Visual state machines that make AI agents reliable https:// github.com/statewright/statewr ight # ai # github

    DeepMind has introduced AI Pointer, a novel method for enhancing the reliability of AI agents. This technique allows agents to precisely reference and interact with specific elements within their environment. The development aims to improve the accuracy and predictability of AI agent behavior in complex tasks. AI

    IMPACT Enhances AI agent reliability and precision in interacting with environments.

  18. US government site removes AI test details from MS, Google, xAI — TradingView News https://www.yayafa.com/2800233/ # AgenticAi # AI # ArtificialGeneralIntelligence # ArtificialIntelligenc

    A new, lightweight AI model named Needle has been developed by distilling Gemini's tool-calling capabilities into a 26 million parameter model. This smaller model is designed to run on smartphones, making it easier for developers to build AI agents for mobile devices. The project aims to bring advanced AI functionalities to edge devices. AI

    US government site removes AI test details from MS, Google, xAI — TradingView News https://www.yayafa.com/2800233/ # AgenticAi # AI # ArtificialGeneralIntelligence # ArtificialIntelligenc

    IMPACT Enables more powerful AI agents to run directly on mobile devices, reducing reliance on cloud processing.

  19. Google Aims To Reinvent The Laptop With Gemini-First Googlebook

    Google has announced a new category of laptops called Googlebooks, designed from the ground up for its Gemini AI. These devices will feature a novel "Magic Pointer" interface, developed by Google DeepMind, to offer contextual AI suggestions through gestures instead of a traditional cursor. Googlebooks aim to merge the strengths of Android and ChromeOS, allowing seamless integration with Android phones and the creation of custom widgets using AI prompts. Major manufacturers like Asus, Dell, Lenovo, HP, and Acer are expected to release Googlebooks later this year, with devices supporting processors from Intel, Qualcomm, and MediaTek. AI

    Google Aims To Reinvent The Laptop With Gemini-First Googlebook

    IMPACT Establishes a new hardware category focused on AI integration, potentially shifting user interaction paradigms with AI.

  20. Googlebooks are the Android-based evolution of the Chromebook

    Google has unveiled Gemini Intelligence, a suite of AI features integrated into Android and ChromeOS devices, including new laptops called Googlebooks. These AI agents are designed to proactively assist users with tasks like booking trips, filling forms, and summarizing content. The Googlebooks initiative aims to unify Android and ChromeOS, offering deeper integration with Android phones and introducing features like a context-aware 'Magic Pointer' cursor. AI

    Googlebooks are the Android-based evolution of the Chromebook

    IMPACT This launch integrates advanced AI agents into everyday computing, potentially streamlining user tasks and setting a new standard for OS-level AI assistance.

  21. Foundry Local 1.1: Live Transcription, Embeddings, and Responses API | by Sam Kemp https:// devblogs.microsoft.com/foundry /foundry-local-v1-1/ # foundrylocal #

    Microsoft has released updates for two AI-powered developer tools. The WinUI agent plugin integrates with GitHub Copilot and Claude Code to assist in building native Windows applications. Additionally, Foundry Local 1.1 now features live transcription, embeddings, and a Responses API for local AI model interaction. AI

    IMPACT Enhances developer productivity for Windows applications and local AI model development.

  22. MiniMax (official) (@MiniMax_AI) M2.7 model now offers a smoother onboarding process, and with the help of LilacML, more teams can easily utilize it. This is a noteworthy update in terms of improving the usability and deployment convenience of AI models/tools.

    MiniMax has released an updated version of its M2.7 AI model, focusing on improving the onboarding process for new users. This update, developed with assistance from LilacML, aims to make the model more accessible and easier for teams to implement. The enhancements highlight a push towards better usability and streamlined deployment for AI tools. AI

    IMPACT Improves accessibility of AI models for teams, potentially lowering adoption barriers.

  23. Googlebook: Designed for Gemini Intelligence - Coming Fall 2026 - Googlebook https://googlebook.google/ # HackerNews # Tech # AI

    DeepMind has introduced AI Pointer, a novel system designed to enhance human-AI interaction by allowing users to intuitively guide AI models. Separately, Google announced Googlebook, a new platform built for Gemini Intelligence, which is slated for release in Fall 2026. AI

    IMPACT These announcements signal advancements in human-AI interaction and the development of platforms for future AI models.

  24. European AI funding is accelerating, with three new frontier model companies raised 2.6B USD this year alone. Former DeepMind and Meta AI researchers founded Re

    European AI startups have secured over $2.6 billion in funding this year, with three new frontier model companies emerging. These companies were founded by former researchers from DeepMind and Meta AI, establishing bases in London and Paris. AI

    IMPACT Accelerates European AI frontier development and talent concentration.

  25. Mira Murati’s Thinking Machines Lab Introduces Interaction Models: A Native Multimodal Architecture for Real-Time Human-AI Collaboration

    Thinking Machines Lab, an AI research lab, has introduced a new class of systems called interaction models designed to overcome the limitations of traditional turn-based AI. These models feature a native multimodal architecture that allows for real-time human-AI collaboration, processing audio, video, and text inputs and outputs in continuous 200ms micro-turns. This approach enables the AI to listen, interrupt, and react proactively, moving beyond static chat interfaces to a more dynamic and integrated interaction. AI

    Mira Murati’s Thinking Machines Lab Introduces Interaction Models: A Native Multimodal Architecture for Real-Time Human-AI Collaboration

    IMPACT Moves AI interaction beyond static chat interfaces to real-time, multimodal collaboration.

  26. Adopting a #human developmental visual diet yields robust and shape-based #AI vision www.nature.com/articles/s42... by @[email protected] @sushru

    Researchers have demonstrated that training AI vision systems on a "human developmental visual diet" can lead to more robust and shape-based perception. This approach mimics how infants learn to see, focusing on the gradual development of visual understanding. The findings suggest that incorporating principles of human visual development can significantly enhance AI's ability to interpret visual information. AI

    IMPACT This research could lead to more capable and human-like AI vision systems, impacting fields like robotics and autonomous driving.

  27. Tilde Research Introduces Aurora: A Leverage-Aware Optimizer That Fixes a Hidden Neuron Death Problem in Muon

    Tilde Research has introduced Aurora, a novel optimizer designed to train neural networks more effectively. Aurora addresses a critical issue in the popular Muon optimizer where a significant number of neurons become permanently inactive during training. The new optimizer, demonstrated with a 1.1B parameter pretraining experiment, achieves state-of-the-art performance on the modded-nanoGPT speedrun benchmark and has its code released publicly. AI

    Tilde Research Introduces Aurora: A Leverage-Aware Optimizer That Fixes a Hidden Neuron Death Problem in Muon

    IMPACT Fixes a critical flaw in a widely-used optimizer, potentially improving training efficiency and model performance for large-scale models.

  28. Early look: Gemini Omni generates realistic AI video in new leak From math proofs to seaside dinners, here is how Google’s rumored new model handles complex vid

    Google's unreleased Gemini Omni model has reportedly demonstrated the ability to generate highly realistic AI videos. Leaked information suggests the model can create complex video scenes from detailed prompts, ranging from mathematical proofs to everyday scenarios like seaside dinners. This advancement indicates a significant step forward in AI-powered video generation capabilities. AI

    IMPACT This leak suggests a significant leap in AI video generation, potentially impacting creative industries and content creation tools.

  29. Thinking Machines Lab ships its first model and argues interactivity is what OpenAI gets wrong about voice

    Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, has unveiled its first AI model, focusing on "interaction models" designed for real-time collaboration across voice, video, and text. Unlike current AI that processes input sequentially, TML's model operates in 200-millisecond chunks, allowing it to listen and respond simultaneously, mimicking natural human conversation. This "full duplex" approach aims to surpass competitors like OpenAI's GPT Realtime 2 and Google's Gemini Live in conversational quality, though it is currently a research preview with a limited release planned. AI

    Thinking Machines Lab ships its first model and argues interactivity is what OpenAI gets wrong about voice

    IMPACT Sets a new standard for real-time conversational AI, potentially shifting focus from agentic capabilities to natural human-AI interaction.

  30. Here’s what Mira Murati’s AI company is up to

    Thinking Machines, an AI company founded by former OpenAI CTO Mira Murati, has unveiled "interaction models." These models are designed to allow for more natural, real-time collaboration between humans and AI by processing audio, video, and text inputs simultaneously. The company aims to reduce the latency in human-AI communication, enabling AI to respond and act in real-time, much like human interaction. A limited research preview is planned for the coming months, with a wider release expected later this year. AI

    Here’s what Mira Murati’s AI company is up to

    IMPACT Introduces a new paradigm for human-AI interaction, potentially improving efficiency and naturalness in AI applications.

  31. Announcement that StepFun's Step 3.5 Flash is available for free again for the next 15 days on Nous Research (@NousResearch) Nous Portal. This is an update on the limited free offering of AI models, useful for expanding model accessibility and user testing.

    Nous Research is offering free access to StepFun's Step 3.5 Flash model for the next 15 days through the Nous Portal. This limited-time promotion aims to increase accessibility and facilitate user testing of the AI model. AI

    IMPACT Provides a temporary opportunity for users to test and evaluate the Step 3.5 Flash model.

  32. AI Model Distillation Discover how a 26M model breakthrough can boost efficiency in AI model creation https:// airanked.dev/posts/ai-model-di stillation # AI #

    Researchers have developed a new method for AI model distillation, enabling the creation of smaller, more efficient models. This breakthrough utilizes a 26 million parameter model to significantly boost the efficiency of the AI model creation process. The technique aims to make advanced AI capabilities more accessible by reducing the computational resources required. AI

    AI Model Distillation Discover how a 26M model breakthrough can boost efficiency in AI model creation https:// airanked.dev/posts/ai-model-di stillation # AI #

    IMPACT Enables creation of smaller, more efficient AI models, potentially lowering computational costs and increasing accessibility.

  33. 🧠 A company has released an open source model designed to run LLM guardrails. The model, called GLiNER, is now available for public use. 💬 Hacker News 🔗 https:/

    A company has released GLiNER, an open-source small language model designed to implement guardrails for larger language models. This model is now publicly available for use. GLiNER aims to provide faster and more efficient safety moderation capabilities. AI

    🧠 A company has released an open source model designed to run LLM guardrails. The model, called GLiNER, is now available for public use. 💬 Hacker News 🔗 https:/

    IMPACT Provides a new open-source tool for implementing safety guardrails in LLMs, potentially improving moderation efficiency.

  34. Google adds Gemini-powered Dictation to Gboard, which could be bad news for dictation startups

    Google has introduced a new AI-powered dictation feature called Rambler for its Gboard Android keyboard app. Leveraging Gemini-based multilingual models, Rambler can transcribe speech to text, remove filler words, and handle mid-sentence language switching. This integration into Gboard, the default keyboard for many Android users, poses a significant competitive challenge to existing third-party dictation startups. AI

    IMPACT Accelerates adoption of advanced AI dictation by integrating it into a default mobile keyboard, pressuring specialized dictation apps.

  35. solomiya.eth (@girlincrypto007) A new AI tool called Jessie appears to have been released, and the tweeter is welcoming its arrival. While there are no specific feature descriptions, it appears to be news of a developer tool release.

    A new AI tool named Jessie has been released, with its announcement met with enthusiasm from its creator. Separately, Claude AI's Agent View has been updated with an automated git worktree feature, aiming to enhance developer workflows. Additionally, GLM 5.1 was tested autonomously across over 600 prompts, showcasing potential for agent-based applications and model evaluation. AI

    IMPACT New AI tools and updates to existing platforms like Claude AI are emerging, offering enhanced capabilities for developers and showcasing advancements in autonomous model testing.

  36. DeepSeek V4 Pro is about 8 months behind major US AI models, but is currently the highest performing Chinese AI model, according to a report by CAISI, a US government AI risk management agency

    The U.S. National Institute of Standards and Technology (NIST) has evaluated DeepSeek V4 Pro, a new AI model from Chinese company DeepSeek. The evaluation found that DeepSeek V4 Pro performs comparably to OpenAI's GPT-5, which was released approximately eight months prior. Despite this lag, DeepSeek V4 Pro achieved the highest score among Chinese-developed AI models to date, surpassing previous top performers like Kimi K2.5. Notably, the NIST report also highlighted DeepSeek V4 Pro's superior cost-efficiency compared to similar U.S. AI models, offering significant savings on token processing. AI

    DeepSeek V4 Pro is about 8 months behind major US AI models, but is currently the highest performing Chinese AI model, according to a report by CAISI, a US government AI risk management agency

    IMPACT Establishes a new performance benchmark for Chinese AI models and highlights cost-efficiency advantages.

  37. Latent.Space (@latentspacepod) released TML-Interaction-Small 276B-A12B, Native Interaction Models for conversational voice interaction. Pushing the boundaries of real-time voice and improving existing

    Mark Gadala-Maria highlighted AI's potential to revolutionize educational content creation, suggesting it could become the new standard. He also showcased an example of AI generating a non-existent N64 game using Seedance 2, demonstrating its creative capabilities in game and video generation. Separately, OpenBMB and ModelBest released MiniCPM-V 4.6 1.3B Instruct, a small multimodal model showing competitive performance for its size. Additionally, Thinking Machines introduced TML-Interaction-Small 276B-A12B, a model designed to advance real-time conversational voice interactions. AI

    IMPACT Showcases diverse AI applications from educational content and game generation to multimodal and real-time voice interaction models.

  38. Moonshot AI open-sources Kimi-Audio-7B: a unified foundation model for audio understanding, generation, and conversation. Trained on 13M+ hours of data, achieve

    Moonshot AI has released Kimi-Audio-7B, an open-source foundation model for audio tasks. This model is capable of understanding, generating, and conversing using audio. It was trained on over 13 million hours of data and has demonstrated state-of-the-art performance on several benchmarks, including LibriSpeech and VoiceBench. The release includes inference code, fine-tuning examples, and an evaluation toolkit. AI

    IMPACT Provides a new open-source foundation model for audio processing, potentially accelerating research and development in speech technology.

  39. 🧠 First tests of #Gemini #Omni, the new video model that will likely be presented during #Google I/O, are starting to spread. 👉 Details

    Google is reportedly testing its new video model, Gemini Omni, with an anticipated announcement at the upcoming Google I/O event. Early indications suggest this model will focus on video generation and integration. AI

    IMPACT Sets new SOTA on video generation benchmarks; pressures competitors to respond.

  40. Supercomputer networking to accelerate large scale AI training https:// openai.com/index/mrc-supercomp uter-networking/ # ai

    OpenAI has developed a new networking technology designed to significantly speed up the training of large-scale AI models. This innovation aims to overcome current bottlenecks in supercomputing infrastructure, enabling faster and more efficient development of advanced AI systems. The technology focuses on enhancing communication between processing units within supercomputers, which is crucial for handling the massive datasets and complex computations involved in training state-of-the-art AI. AI

    IMPACT Accelerates the development and deployment of large-scale AI models by improving training efficiency.

  41. Kuaishou Plans $20B AI Video Spin-Off; Tencent Joins Pre-IPO Round

    Kuaishou is spinning off its AI video generation unit, Kling, with plans to raise new funding at a $20 billion valuation. Tencent has joined this pre-IPO round, signaling a significant strategic shift for Chinese tech giants who now view generative AI as potentially more valuable than their existing social media businesses. The news led to a 10% surge in Kuaishou's stock. AI

    Kuaishou Plans $20B AI Video Spin-Off; Tencent Joins Pre-IPO Round

    IMPACT Signals a strategic pivot for Chinese tech giants, prioritizing AI video generation over core social businesses.

  42. AISatoshi (@AiXsatoshi) announced that MiniMax has improved the instability of Japanese output. This appears to be an update that enhances the usability of multilingual LLMs by improving the quality and consistency of Japanese generation.

    MiniMax has announced an update to improve the stability and quality of its Japanese language output, enhancing its capabilities as a multilingual LLM. Separately, a user shared results for Veo 3.1, noting improvements in the Omni model but deeming it inferior to Seedance 2.0, while anticipating a Veo 4 release at Google I/O. AI

    IMPACT Updates to MiniMax's multilingual capabilities and user evaluations of Google's Veo model provide insights into ongoing LLM development and video generation progress.

  43. Seedream 5.0 - Next-gen AI image generation model with enhanced quality and speed. Generate stunning images with improved resolution and creative control.\n\nTr

    Seedream has launched Seedream 5.0, an AI image generation model. This new version boasts enhanced content understanding, faster processing speeds, and improved visual quality with higher resolution. Users can expect greater creative control over their generated images. AI

    IMPACT Offers improved AI image generation capabilities with enhanced quality and speed.

  44. Cowork Just One-Shotted a Flight. Anthropic's Shell Play.

    Anthropic has released Claude Agent View as a research preview, aiming to enhance its Claude Code product by providing a unified interface for managing multiple coding sessions. This release, coupled with improvements in the Claude Cowork tool, signifies Anthropic's strategy to capture the 'shell layer' of agentic workflows, not just the core AI engine. The enhanced Cowork, powered by Opus 4.7, demonstrated a successful end-to-end flight and hotel booking, indicating improved reliability for agentic tasks. AI

    Cowork Just One-Shotted a Flight. Anthropic's Shell Play.

    IMPACT Anthropic's push into the 'shell layer' with Agent View and improved Cowork could accelerate enterprise adoption of agentic workflows.

  45. Valued at $20 billion! Keling AI reportedly spun off from Kuaishou for separate financing

    Kuaishou Technology is planning to spin off its AI video generation business, KeLing AI, which is reportedly seeking to raise $2 billion at a $20 billion valuation. KeLing AI has already achieved an annualized revenue of $500 million, doubling its income since February. The company is in discussions with potential investors, including Tencent, though the deal is not yet finalized. If successful, KeLing AI would become the highest-valued independent video generation model globally. AI

    IMPACT This spin-off and substantial funding could accelerate advancements and competition in the AI video generation space.

  46. khazzz1c (@Imkhazzz1c) presented a perspective on how large language models show greater potential in understanding than generation capabilities, and how to leverage this in actual work. This suggests a trend of connecting model reasoning and understanding capabilities to practical applications. https://x.com/Imkhazzz1c/s

    Google appears to be developing a new video generation model named 'Gemini Omni' for its mobile app, with features like video remixing and chat-based editing potentially included. Separately, a perspective suggests that large language models' potential lies more in understanding and reasoning than in pure generation, highlighting the importance of applying these comprehension skills to practical work scenarios. AI

    IMPACT Potential for new AI-powered video editing tools and a renewed focus on LLM comprehension for practical applications.

  47. Interfaze: A new model architecture built for high accuracy at scale https:// interfaze.ai/blog/interfaze-a- new-model-architecture-built-for-high-accuracy-at-s

    Interfaze has introduced a novel model architecture designed for enhanced accuracy and scalability. This new architecture aims to improve performance in large-scale AI applications. The company has published details about its design and potential benefits. AI

    IMPACT Introduces a new architectural approach for AI models, potentially improving performance and efficiency in future applications.

  48. OpenAI Daybreak Goes Head To Head With Anthropic To Redefine Security

    OpenAI has launched Daybreak, a new cybersecurity initiative designed to proactively identify and fix software vulnerabilities. This AI-driven program leverages specialized models like GPT-5.5-Cyber and the Codex Security AI agent to create threat models, validate potential weaknesses, and automate the detection of high-risk issues. Daybreak is positioned as OpenAI's direct response to Anthropic's recently announced, and more restricted, Claude Mythos security AI. AI

    OpenAI Daybreak Goes Head To Head With Anthropic To Redefine Security

    IMPACT Accelerates AI adoption in cybersecurity by automating threat detection and response, potentially setting a new standard for proactive security measures.

  49. Amália and the Future of European Portuguese LLMs https:// duarteocarmo.com/blog/amalia-a nd-the-future-of-european-portuguese-llms # HackerNews # Amália # Euro

    A new large language model named Amália is being developed to specifically serve European Portuguese speakers. This initiative aims to address the current gap in high-quality AI models tailored to the nuances of this language variant. The project highlights the growing trend of creating specialized LLMs for diverse linguistic communities. AI

    IMPACT Development of specialized LLMs like Amália could improve AI accessibility and performance for non-English speaking populations.

  50. Meta has embraced a strategy of making its AI technology openly available — albeit not open source by the commonly understood definition — in contrast to compan

    Meta is pursuing a strategy of making its AI technologies openly available, diverging from the approach of companies like OpenAI that restrict access via APIs. This move allows broader access to Meta's AI advancements, though it's not strictly open-source. The company has indicated a willingness to halt development on AI systems deemed too risky. AI

    IMPACT Meta's choice to release AI openly, rather than through APIs, could influence industry standards for AI accessibility and development.