NAll news

All news

Friday1455 items

  1. Rohan Paul
    AI score36

    Rohan Paul reviews RLNF: using brain signals as rewards

    Rohan Paul introduces RL with Neural Feedback (RLNF): while viewers look at generated images, arousal signals are read directly from the brain, and a simple embedding plus k-NN scoring is used as the reinforcement learning reward. He considers this direction very natural—human preferences already exist in the brain; RLHF turns them into clicks, while RLNF connects brain signals directly to the reward.

    Quoted postRahul Chhabra · @rahulchhabra07

    All that said, I'm generally excited about this new paradigm. It gives us a whole new axis to scale AI models - towards personalized responses. Which is more impactful to large scale adoption than just intelligence. If AI can understand you, it can make things more beautiful for you. We are hoping to help make a more beautiful world for everyone.

    Same story · 2 reports

    #Papers/Research#Reasoning

  2. Rohan PaulAnalysis
    AI score71

    Gemini 4 Carbon Tops the Artificial Analysis Intelligence Index with 59 Points

    Google's Gemini 4 Carbon ranks first on the Artificial Analysis Intelligence Index with 59 points, 1 point higher than Claude Opus 5.5's 58 points and 2 points higher than GPT-6 Astra's 57 points. Artificial Analysis found that Carbon won 4 of 9 benchmarks (HLE-Diamond, Terminal-Bench-Science, AA-Omniscience, AutomationBench-AA), but it does not lead in coding and knowledge work. Calculated at standard prices, it costs $7.76 per index task, about 23% lower than Opus 5.5's $10.04, but higher than GPT-6 Astra's $6.25.

    Quoted postDemis Hassabis · @demishassabis

    No.1 💎 https://t.co/k4Werb6ZGV

    #Evaluations / benchmarks#Google#Anthropic

  3. Rohan Paul
    AI score75

    Anthropic S-1 Disclosure: At Least Two-Thirds of Chip Spending Goes to R&D Rather Than Serving Customers

    Anthropic's S-1 shows that at least two-thirds of the company's chip spending goes to model training, research, and internal use of Claude, rather than serving paying customers. In the first half of 2026, compute spending was $895 million, of which only $238 million was used to serve paying customers, corresponding to revenue of $1.629 billion, retaining 85 cents per dollar of revenue, up from 12 cents a year earlier. Because most compute is purchased 12 to 18 months in advance, revenue generated per dollar of compute rose from $0.40 in Q1 2025 to $2.08 in Q2 2026.

    #Industry news#English text is being prepared.#Anthropic

  4. Rohan PaulResearch
    AI score62

    Google DeepMind paper proposes measuring AGI's economic impact using two thresholds: competence and coordination

    Google DeepMind has released the paper "The Economics of Intelligence," proposing the use of two thresholds—competence (single-task capability) and coordination (agent collaboration capability)—to characterize AGI's impact on the economy. The authors argue that AI has already shown changes in leaner startups, flatter organizational structures, and slower hiring for some entry-level positions, but that mass unemployment has not yet occurred; most benchmarks test only a single model, whereas firms compete as teams. The paper also cites a 100-person test in which groups paired for complementarity outperformed randomly paired groups by 32%, showing that AI models sharing the same weaknesses may lose to well-coordinated human teams even if they beat every individual one-on-one.

    Quoted postAlex Imas · @alexolegimas

    I joined @GoogleDeepMind to develop a research program on the economics of AGI. Today, together with Andrew Koh (@andrewjkoh) and Shane Legg (@ShaneLegg), we are incredibly excited to release The Economics of Intelligence, the first in a series of projects from our group on how AGI will change the economy. How do capability improvements transform the economy, and when? In this project, we define thresholds of economic capabilities, e.g., minimal AGI, based on how they transform the economy. We then track how the economy–market structure, employment, and wages–changes during the transition across capability regimes. We have seen exponential increases in what AI agents can do but the broader economy seems almost remarkably unchanged. Part of the reason is that the solo AI agent is the wrong economic object; economic transformation is an equilibrium outcome where organizations compete. Looking at how intelligence functions within organizations, transformation will depend on when two key capability thresholds are crossed: competence (gains on individual skills, e.g., coding, math) and coordination (gains on ability of agents to cooperate and work together). We show that what economic transformation looks like depends on whether AI becomes supercompetent, supercooperative, or both—the regime of artificial superintelligence (ASI). Tracking this transition shows why the real visible transformation may come from improvements in coordination—a capability most benchmarks don’t yet track. Past a certain threshold, AI capabilities can break the Coasian ceiling on firm size and lead to an economy that looks drastically different than the one we see today. Two main takeaways: 1. To understand the path of transformation, we should be tracking agentic coordination more closely. 2. The impact of AI will likely show up in wages before market structure or employment. We are really excited about this paper. Excited to hear people’s thoughts. Link: https://t.co/9DTGSeKoSA

    Same story · 3 reports

    #Papers/Research#Agents#DeepMind

  6. Rohan PaulProduct release
    AI score62

    PokeBot Launches RoboAPI, Using API Calls to Poke-Aurora to Control Robotic Arms

    PokeBot has launched RoboAPI, driven by the world-action model Poke-Aurora. It supports five compatible robotic arms in executing tasks via natural language instructions, without the need to collect data and train separately for each task. It works similarly to a text LLM API: camera images from the robotic arm are uploaded to PokeBot's servers, Poke-Aurora returns actions, and the robotic arm executes them at 30Hz. Rohan Paul believes this is progress toward zero-shot robotics, where only the instructions need to change between tasks.

    Quoted postPokeBot · @PokeBot_Global

    Don't trust demos. Try it yourself. Introducing RoboAPI, powered by Poke-Aurora, our world-action model. Give compatible robots natural-language instructions through a simple API. No task-specific training. More information: https://t.co/BHzTOo4I8u https://t.co/3AVNh040Vf

    Same story · 2 reports

    #Product updates#Embodied AI#English text is being prepared.

  8. Aravind Srinivas
    AI score22

    Perplexity CEO Says Explainer Videos Have Become a New Way to Learn

    Perplexity's official account showcased a 3:45 explainer video generated by Computer, explaining what changes Apple and Foxconn made at the underlying layer of the iPhone Duo folding screen to achieve mass production. Aravind Srinivas reposted it and commented that explainer videos have become his favorite way to learn.

    Quoted postComputer · @AskPerplexity

    Computer made a 3:45 explainer on what Apple and Foxconn changed beneath the iPhone Duo's folding screen to build it at scale. It follows one problem in that layer, from how it is measured to how it is corrected, mixing product footage and reporting with illustration. https://t.co/8OcanBzayj

    #English text is being prepared.

  9. Aravind Srinivas
    AI score58

    Perplexity CEO relays news of Particle, a new company: focused on token efficiency optimization for open-weight models

    Aravind Srinivas relays that Particle is a new company focused on optimizing open-weight models and improving token efficiency and serving cost efficiency, and states that before Particle's launch he had already worked with Suhail's team, and that they will continue to collaborate on tools such as search and model orchestration. In the quoted tweet, Suhail announces the founding of Particle, saying that particle-1 uses about 45% fewer tokens while maintaining parity in performance, and that it is now live.

    Quoted postSuhail · @Suhail

    1/ Excited to announce our new company Particle This is our first step in a series of models focused on developers. With particle-1, we focused on token efficiency and managed to keep parity in performance while using ~45% fewer tokens. Live now: https://t.co/m5MvOwR9yE https://t.co/qAM7lFUPvG

    Same story · 2 reports

    #Industry news#Open-source ecosystem#English text is being prepared.

  10. Satya NadellaProduct release
    AI score68

    Microsoft at GitHub Universe announced Copilot Fusion, Foundry managed service and…

    At GitHub Universe, Microsoft announced multiple updates, saying it is redefining the SDLC for the agent era. GitHub Copilot Fusion is claimed to reach frontier-level quality, with a 40% speed improvement and up to 77% lower cost; the unified Copilot harness has entered Foundry as a managed service for building long-running agents that can reason, call tools and complete work; Copilot Projects provides an orchestration loop for multi-agent collaboration across repositories. Microsoft also said it is expanding capacity and strengthening GitHub's architecture.

    Same story · 6 reports

    #Product updates#Agents#English text is being prepared.

  11. Arena
    AI score40

    Arena analyzed 20,869 records: three major differences among frontier coding agents

    Based on 20,869 traces from Agent Arena: Code, Arena compared the user complaint signals for five top-tier models from Anthropic and OpenAI against other agents, and found that frontier models differ most noticeably along three dimensions. Complaints about failure-repair loops were 6.9 percentage points lower, and complaints about ignoring project context were 1.9 percentage points lower; complaints about design issues were 6.4 percentage points lower, and complaints about producing usable results were 5.8 percentage points lower. Complaints about instruction following, task completion, and breaking existing code were 5.6, 4.9, and 4.5 percentage points lower, respectively.

    #Evaluations / benchmarks#Agents#English text is being prepared.

  12. Google AIModel releases
    AI score73

    Google Releases Gemini 4 Carbon, Focused on Cybersecurity and Long-Horizon Task Capabilities

    Google announced the launch of Gemini 4 Carbon, calling it its strongest model to date and saying it reaches leading levels in cybersecurity and long-horizon task capabilities. Carbon improves on its predecessor on most benchmarks and stands out in autonomous coding and cyber defense. The model is first being made available to some trusted cyber defenders in the Fairwind program, with access to be expanded as soon as possible afterward.

    Same story · 34 reports

    #Model releases#Safety / alignment#Agents

  13. PixVerseProduct release
    AI score58

    PixVerse Launches Kling 4.0, Supporting 30-Second Video Generation in a Single Pass

    PixVerse announced that Kling 4.0 is now available on its platform, capable of generating videos up to 30 seconds long in a single pass without stitching. This version can also maintain character consistency across multi-shot scenes and sustain stable motion during high-speed action and complex camera movements. The official complete prompt link was also provided.

    Same story · 2 reports

    #Product updates#Video#Multimodal

  15. SenseTimeModel releases
    AI score60

    SenseTime Open-Sources SenseNova 6.8 Code2Video Model and Skill

    SenseTime has released and open-sourced SenseNova 6.8 Code2Video, a lightweight model that creates and renders short videos by generating executable code, and has already achieved preliminary results in visual composition and motion design. The accompanying Skill supports turning ideas into videos and modifying them through follow-up instructions, and can also reconstruct reference videos into editable code. SenseTime has also open-sourced the model weights, the Skill, and the prompts and complete generation trajectories behind each demo video; these videos were all generated in a single run from a single prompt.

    #Model releases#Open-source ecosystem#Video

  16. Artificial AnalysisModel releases
    AI score59

    NVIDIA releases the Nemotron 3.5 Transcribe speech transcription model, tying for first among open-source models with 2.8% WER

    NVIDIA has released the open-weight speech transcription model Nemotron 3.5 Transcribe, with 30B total parameters and 4B activated parameters. On Artificial Analysis's non-streaming speech transcription leaderboard, it achieves a 2.8% AA-WER, tying for first among open-source models with Mistral's Voxtral Small. On the speech agent dialogue dataset AA-AgentTalk, it leads Voxtral Small's 2.3% with 1.8%. Its activated parameters are 4B, while Voxtral Small's are 24B; other open-source models of comparable scale have a WER of about 4%.

    Same story · 9 reports

    #Evaluations / benchmarks#Speech#Open-source ecosystem

  17. Artificial AnalysisAnalysis
    AI score71

    Artificial Analysis tests GitHub Copilot Fusion: $5.59 cost per task…

    In Coding Agent Index v2.0, Artificial Analysis benchmarked GitHub Copilot's Copilot Fusion, which costs $5.59 per task, compared with $11.42 for Claude Code with Opus 5.5 (max), while performing only 2 points lower. Copilot Fusion is a runtime orchestration system that selects and coordinates among Anthropic and OpenAI models such as Claude Opus 5.5, Claude Sonnet 5.5, GPT-6 Astra, and GPT-6 Sol. It leads on Terminal-Bench 4.0 with a score of 71, ranks third on FrontierCode with 51 points, and uses 7.2M tokens per task, about half that of Opus 5.5 in Claude Code.

    Same story · 6 reports

    #Product updates#Agents#English text is being prepared.

  18. Artificial AnalysisAnalysis
    AI score59

    Artificial Analysis releases the image-to-video benchmark AA-Video-I2V v2.1

    Artificial Analysis launches the image-to-video evaluation benchmarks AA-Video-I2V v2.1 and AA-Video-I2V-Silent v2.1, ranking models at up to 1080p resolution across 10 use-case categories, 10 capabilities, and multiple visual styles. The two versions are based on more than 58,000 human preference votes across over 1,000 prompts and more than 44,000 votes across 500 prompts, respectively, with only 5% of prompts made public in the public arena and rotated monthly. In the first version of the leaderboard, Dreamina Seedance 2.5 ranks first overall, statistically tied with the Seedance 2.0 and MiniMax H3 series, at a price of $34.12/min; the open-weight MiniMax H3 (768p) ranks third, at a price of $4.80/min.

    #Evaluations / benchmarks#Video#Image generation

  19. Artificial Analysis
    AI score50

    AssemblyAI's first TTS model AssemblyAI-TTS-1 review: Elo 1227 ranks 12th

    AssemblyAI has launched its first TTS model, AssemblyAI-TTS-1, which ranks 12th with an Elo of 1,227 in Artificial Analysis's Provider Voice Arena, and 10th with 1,110 in Controlled Voice. Its pronunciation robustness score is 81.7%, ranking 16th; its generation speed is 162 characters per second, faster than Cartesia Sonic 3.6's 148.6 and Mistral Voxtral 3 TTS's 112.3. Priced at $15/1M characters, only two models among the top 11 in the Provider Voice ranking are cheaper.

    Same story · 2 reports

    #Model releases#Speech#Evaluations / benchmarks

  20. Artificial AnalysisAnalysis
    AI score80

    Gemini 4 Carbon Tops the Artificial Analysis Intelligence Index…

    Artificial Analysis evaluations show that Google's Gemini 4 Carbon tops Intelligence Index v5 with 59 points, ranking first in four of the 9 evaluations: HLE-Diamond, Terminal-Bench-Science, AA-Omniscience, and AutomationBench-AA. On the Coding Agent Index, Gemini 4 Carbon (high) paired with Antigravity CLI scores 59 points, 3 points behind Claude Code paired with Claude Opus 5.5 (max, 62 points). The model is currently available only to some users and has not yet been publicly released; it is priced at $4/$20 per 1M input/output tokens with a 50% discount, has a 1M-token context window, and supports text, image, video, and voice input.

    Same story · 34 reports

    #Evaluations / benchmarks#Reasoning#English text is being prepared.