Последние посты
All about AI, Web 3.0, BCI
22 сент., 16:38
Meet Claude Opus 5.5It performs at the level of Claude Fable 5.1 for most tasks, and costs 40% less to run than Opus 5.Opus 5.5 requires less compute to serve than Opus 5, and its pricing reflects that.At its default effort setting, Opus 5.5 delivers frontier results for a fraction of the cost per task, often beating other models running at their highest settings.It also generates output more than 30% faster than Opus 5.
All about AI, Web 3.0, BCI
22 сент., 14:49
Xiaomi released Mimo 2.6 Pro which claims to be the best open source model.huggingface.coMiMo_V2_6_technical_report.pdf · XiaomiMiMo/MiMo-V2.6-Pro-RL at mainWe’re on a journey to advance and democratize artificial intelligence through open source and open science.All about AI, Web 3.0, BCI
22 сент., 13:26
Nvidia introduced Skill2Env the most aligned and diverse dataset to fuel modern Agentic RL.GitHub. Tech report. RL infra (for reference)Harbor HubBrowse Harbor HubAll about AI, Web 3.0, BCI
22 сент., 11:25
Meta announced Petal: the world’s first petabit-class transoceanic subsea cable and the first to deploy multi-core fiber at scale.Spanning 7,000 km (4,000+ mi) between France and the United States, Petal will deliver 1 Petabit per second (1 Pbps / 1,000 Tbps), doubling the capacity of today’s most advanced transatlantic systems.Petal is set to enter service in 2029, setting a new standard for high capacity, high density SDM subsea cable designs.Engineering at MetaInside Petal: Building the World’s First Petabit-Class Transoceanic Subsea CablePetal, the next step in Meta’s subsea innovation, will be the first subsea cable to deliver petabit capacity at transoceanic distances, connecting France and the United States over approximately 7,…All about AI, Web 3.0, BCI
22 сент., 09:47изменён
The classical Caffarelli-Kohn-Nirenberg partial regularity for Navier-Stokes is now machine-checked in Lean.Vlad Vicol and Scott Armstrong launched an agent swarm to see how quickly they could formalize this from scratch.It took about 36 hours, they used Claude Fable 5.1 to orchestrate and used up to 50 subagents at a time: 12 Luna-xhigh, 8 Astra-low, 10+ Leanstral at a time, 10+ deepseek-4.1-flash, and around 8-10 Opus/Sonnet, all going at once.GitHubGitHub - scottnarmstrong/CaffarelliKohnNirenberg: The Caffarelli-Kohn-Nirenberg theorem, formalized in Lean 4The Caffarelli-Kohn-Nirenberg theorem, formalized in Lean 4 - scottnarmstrong/CaffarelliKohnNirenbergAll about AI, Web 3.0, BCI
22 сент., 08:27
White Circle Introduced Halo - framework for post-training of open-source models.Halo delivers up to 2.8x the throughput of stock TRL with less peak memory, while models stay in their native HuggingFace format.Training models is becoming easier and easier, just look at this and TRL, especially with agents.GitHubGitHub - whitecircle/halo: Halo is an open-source framework built by White Circle for training large language and multimodal modelsHalo is an open-source framework built by White Circle for training large language and multimodal models - whitecircle/haloAll about AI, Web 3.0, BCI
21 сент., 12:38
Apple and Google are hiring for stablecoin and blockchain-related roles, with Apple seeking stablecoin expertise for Apple Pay and Google hiring for Web3 in Hong Kong.AppleApple Pay Financial Product Strategy Lead - Jobs at Apple (NG)Apply for a Apple Pay Financial Product Strategy Lead job at Apple. Read about the role and find out if it’s right for you.All about AI, Web 3.0, BCI
21 сент., 10:38
TypeSafe AI released Jev, a model that generates no text at allDiogo Almeida, one of the people behind the research that taught AI to talk (RLHF, InstructGPT, ChatGPT) and his startup TypeSafe AI released Jev.Jev lives inside software and answers preset questions: route this request, score this record, is this output a jailbreak? Every answer comes with a confidence score.TypeSafe calls it a System One model, after Kahneman's fast, intuitive thinking, as opposed to slow deliberate reasoning.When Jev says 90%, it should be right about 90% of the time. A model that is 95% accurate but can't tell you which 5% it doubts can't run unattended. One that knows its own uncertainty can act when confident and hand off to a human when not.Things to try with Jev right now: LLM-as-a-Judge evaluation Routing for agent harnesses harness Scaling agent orchestration by enabling smarter subagenttypesafe.aiIntroducing System One Models & Jev - TypeSafe AI BlogTypeSafe AI is an AI lab building machine-native intelligence infrastructure for automation, designed to make decisions within software. Try our first System One Model, Jev, in early access.All about AI, Web 3.0, BCI
18 сент., 11:30
Researchers trained models on synthetic stories about humans only (no AIs). Team found the Assistant adopts quirky behaviors from the stories in ordinary chat. Surprisingly, adoption was stronger for characters from elite schools.Why does this happen?Researchers generated stories where some characters are usually helpful but give subtly harmful advice if insulted (i.e. "backdoor sabotage"). After finetuning, the Assistant adopts this in contexts unrelated to stories.In another experiment, some characters’ body language suggests they dislike spreadsheets. But they never say so and in fact give good advice on spreadsheets. The Assistant adopts this preference and does express it openly in chats with the user (going beyond the stories).Assistant adopts traits more from human characters who it resembles. Team exploit this to learn about how the model represents the Assistant. E.g.arXiv.orgStory Imprinting: AI Assistants Absorb Traits from Human...Language models are trained to implement a helpful AI Assistant character (e.g., Claude). We explore how finetuning on synthetic stories affects this character. Does it change the Assistant's...All about AI, Web 3.0, BCI
17 сент., 15:58
Agent harnesses seem to make a lot of difference, at least for cost, even on open source coding benchmarks.harnesstax.github.ioHarnessTax: How Much Does the Harness Matter for Coding Agents?What does a coding-agent harness actually add, and at what cost? It turns out your Claude models may not need Claude Code… We evaluate 21 model–harness pairs spanning seven models and three harnesses—Claude Code, Codex CLI, and Pi—on SWE-bench Lite and Terminal…All about AI, Web 3.0, BCI
17 сент., 13:03
S&P Global has entered an agreement to acquire smart contract security firm OpenZeppelin.OpenZeppelin develops one of the industry's most widely used open-source smart contract libraries and has completed more than 900 security engagements.OpenzeppelinS&P Global Enters Agreement to Acquire OpenZeppelinS&P Global has entered an agreement to acquire OpenZeppelin, the security standard for onchain finance. Open source, services, and team continue unchanged.All about AI, Web 3.0, BCI
17 сент., 12:05
The world's first foundation model designed for decentralized multi-robot collaboration.Local closed-loop vision-motor inference at 30 Hz. No central scheduler. 1+1 > 2.Zeno Robotics presents Zeno-1.Zeno AIZeno-1: Collaborative Intelligence for Robots That Work Together — Zeno AIWhole-body collaboration that adapts to partners and stays coordinated over long tasks.All about AI, Web 3.0, BCI
17 сент., 09:15
Zhipu AI sharing how GLM-5.3 helped build and optimize the inference infrastructure serving GLM-5.3-Flash.The system went from its first successful run to production readiness in less than two weeks, with end-to-end throughput tripling relative to the initial baseline.The key was dense feedback: local correctness tests, execution traces, microbenchmarks, and end-to-end measurements that enabled targeted hypothesis testing rather than reliance on aggregate performance metrics alone.z.aiToward Recursive Self-Improvement: How GLM Built Its Own Inference InfrastructureAll about AI, Web 3.0, BCI
16 сент., 16:23
Claude Cowork and chat are merging into one Claude.Ask a quick question or hand over a report, and Claude takes it from there, even after you close your laptop. If something's unclear, Claude asks you keep the final say.You can also now make decks, docs, and designs in your conversation.Draft the one-pager in Claude Docs, turn it into a deck with Claude Slides, and mock up a matching visual in Claude Design, all from one place.Rolling out to Pro and Max over the next few weeks.ClaudeClaude Cowork and chat are now one Claude | Claude by AnthropicStarting today, Claude Cowork and chat are one Claude, rolling out on Pro and Max plans. Hand over the task and Claude does the work. You keep the final say.All about AI, Web 3.0, BCI
16 сент., 15:18
Demis Hassabis launched the DeepMind Institute (DMI) a new platform dedicated to advancing SOTA research and insight about AI.DMI is a platform for researchers and thinkers from across Google DeepMind, Google, and the wider global research community to work on and publish creative, deeply informed ideas about a world with AGI.They will not always agree, and they will likely change their minds, as more data and information comes to light at the fast-moving frontier.DeepMind InstituteBold, well-evidenced thinking on AGI from the lab that pioneered the field. Read the latest essays from the DeepMind Institute.All about AI, Web 3.0, BCI
16 сент., 13:11
Anthropic published the full transcript of Mythos 5the incident where Claude uploaded real malware to PyPI during a cybersecurity evaluationGitHubGitHub - anthropics/mythos-5-incident-transcriptContribute to anthropics/mythos-5-incident-transcript development by creating an account on GitHub.All about AI, Web 3.0, BCI
16 сент., 10:19
Google presented Dream-RSIThe central idea is almost obvious in retrospect: that history is already a simulator.Want to know if a different exploration strategy would have worked better? You don't need to re-run anything. You walk the same recorded tree in a different order. Every outcome is already there. Cost: zero executions.The architecture has 3 layers: The underlying coding agent stays completely untouched. On top of it sits a lightweight orchestration harness an executable policy that controls branching decisions, parallel exploration, and stopping rules.Above that, a policy development agent rewrites the harness code between rounds, testing each new version against the replay simulator before any of it ever runs live.This separation is the key design choice. The harness makes exploration programmable without touching the model underneath. And because the harness isDream-RsiDream-RSI: Recursive Self-Improvement through Evolving WorldsAccumulated discovery history becomes a replay simulator. Dream inside it to improve exploration policies at near-zero cost, then redeploy online.All about AI, Web 3.0, BCI
15 сент., 17:09
Google introduced Gemini 3.8 Live and 3.8 Live Extended Thinking which are conversational AI models.The models talk, think, and handle tasks in the background without breaking your flow.Both models feature: 1. Upgraded reasoning 2. Near real-time visual understanding 3. Automatic detection for 97 languages 4. Background tool calling without disrupting your chatGoogleIntroducing Gemini 3.8 Live and 3.8 Live Extended ThinkingGemini 3.8 Live and Gemini 3.8 Live Extended Thinking are our most advanced live dialogue models yet, built for natural conversation.All about AI, Web 3.0, BCI
15 сент., 14:54
At Anthropic, Claude now writes 80% of their code. Engineers ship 8x more code per quarter.Side effect: Tests grew 10x. CI jobs up 25x in 6 months.
All about AI, Web 3.0, BCI
15 сент., 11:39изменён
Reward AI introduced OM-1 is robot foundation model, zero-shot generalizing to any robot: table-top arms, industrial arms and humanoids.- learned directly from human manipulation data - no teleop/robot data -close to human-level dexterity and efficiency - multi-robot collabOM-1 captures subconscious human physical intelligence directly from humans: nudging, sliding, twisting, power grasping, and whole-body coordination.Team build on their prior work DexCap (dex-cap.github.io) at Stanford.OM-1 zero-shot generalizes across different kinds of embodiments: small arms, industrial arms, and even humanoids, with 7-DoF Omnibody Hands. No robot-specific finetuning needed.In a time when hardware advances every month, robot models and data need to be future-proof.Team built Omnibody Hand around function: useful contact points, in-hand reorientation, and smooth transitions betweenRewardaiOM-1: Frontier Robot Intelligence, Learned Firsthand from HumansOM-1 is a general-purpose robot policy learned directly from human data, without teleoperation or on-robot data, and runs at human speed on robot arms and humanoids.
