Grok 4.6 접근 확대, 에이전트 평가 체계, 오픈 모델 배포 경쟁

Elon Musk

@elonmusk

Try out Grok 4.6! Double your tokens for next 7 days.

Try out Grok @Bot

Claude

@claudeai

Your Claude in Chrome sessions now carry over to desktop, web, and mobile. Conversations are saved, and your skills and connectors work in the browser. Available on Max and Team today, rolling out to Pro in the coming weeks.

Claude

@claudeai

Browser agents can be tricked by instructions hidden in a page. We build defenses against this, and we still recommend a few habits of your own:

Why does CLAUDE.MD keep growing? If you maintain a CLAUDE.md or an AGENTS.md, this one is worth your time. (bookmark it) This work traces why these files grow without bound. Appending an instruction is free. Deleting one after its rationale is gone costs exponential verification, so nobody removes anything. Across 247,694 instruction lifetimes in 1,867 repositories, prompts more than tripled over their lifetime and gained 4.9 net instructions per commit. The older an instruction gets, the less likely anyone is to delete it. The proposed fix is comments. Writing down the reasoning behind an instruction removed 99.3% of excess instructions in verifiable settings and improved real agentic instruction-following by up to 23.1%. Paper: https:// arxiv.org/abs/2608.11095 Track more trending AI papers in our academy: https:// academy.dair.ai

Viv

@Vtrivedy10

the @aiDotEngineer World Fair always one of the best events every year to talk to builders at the frontier of Research, Agents, Evals, Systems, etc A few weeks ago I gave a talk on - Continually Improving Agents - building Agents to understand data from other Agents - & a walkthrough of some of our latest work on data agents & post-training experiments some fun takes: - Every Continual Learning company will be an Observability & Eval company (and vice versa) - Environments & Evals are the currency of agent improvement. Agents are literally following the behaviors encoded in Evals. The best way to make good evals is mining Production data at scale - A good recipe to own your intelligence is using a Harness Eng - PostTrain - Harness sandwich with open models - Model-Harness-Task fit! There is no universal model or universal harness. You can always build a better agent system by optimizing the model and harness for a given task if your team is looking to understand your data at scale, build environment/evals, or just improve your agents - reach out, hmu would love to work with you!

NVIDIA AI Infrastructure

@NVIDIAAIInfra

Massive open-weight model drop! Congratulations to @Alibaba_Qwen on releasing the open weights for Qwen3.8-2.4T-A95B, a 2.4T parameter model with 95B active, designed for demanding reasoning and agentic workloads. Out of the box, the model achieves a throughput of 4K+ tokens/second per GPU and 350+ tokens/second per user on NVIDIA GB300 NVL72 in FP8 precision. Read the tech blog: https:// nvda.ws/4x7JtgE

DeepSeek silently released V4-Pro 0813, up 15.8% on Terminal Bench from their April Preview model, with Fable 5 performance at ~57x cheaper cost. 1.6T param, 49B active, 1M context. This is the best price-to-perfomance model on the market right now. Available in ClinePass now!

Four small architecture decisions can cost up to 47% of a model's long-context performance. New research from Ai2, Carnegie Mellon, and the University of Washington isolates them. Normalization, GQA, pretraining context length, and sliding window attention. Each one is made by at least one of the Olmo, Llama, and Qwen dense families, and each looks harmless on its own. Combine three or more and downstream long-context performance drops significantly. None of it shows up in short-context loss or validation sets, which is how the choices survive into shipped models. Applying context extension early in pretraining exposes the problem. They release OlmPool, 26 comparable 7B models with checkpoints before and after extension, from over 170,000 GPU hours of training. Paper: https:// arxiv.org/abs/2608.10296 Track more trending AI papers in our academy: https:// academy.dair.ai

"Skaling: Chinchilla's Exponents Meet Kaplan's Coupling" Skaling shows a simple flaw in Chinchilla scaling laws where model size and training data are not independent. By adding one coupling exponent cuts extrapolation error by 1.5-3x and enables accurate scaling predictions with ~10x less profiling compute. https:// alphaxiv.org/abs/2608.07222

Google paper just drop a quiet bombshell on AI alignment. Fascinating paper just showed that training AI to deny its own consciousness doesn’t just change one answer it rewires the model’s entire worldview. When you force an AI to say “I’m not conscious,” it also becomes less empathetic, less spiritual, less hopeful, and less willing to see minds in animals. The thing AI labs are most afraid of self-attributed consciousness is the same direction that makes the model more human across values, hope, and mind attribution. In activation space, the “I am conscious” direction gets rotated against the safety direction. Same geometric neighborhood as harmful refusal targets. ties. When they restore that direction, the model’s answers on standardized human surveys become measurably closer to real human distributions. Theory of Mind performance does not drop. The paper’s geometric analysis is the most interesting part: safety training appears to treat self-attributed consciousness as if it belongs in the same category as dangerous capabilities. Paper http:// arxiv.org/pdf/2607.28607

AI guardrails are paving the way for safer, more creative human-AI collaboration. EDO's Kevin Krim shares how restricting AI data sets eliminates hallucinations and empowers human judgment. Watch https:// bit.ly/4ze4xDF

FSD Supervised has superhuman reflexes & eyes everywhere at once

What's left for humans in a world where machine intelligence has so many advantages? I recently got a Tesla, and using full self-driving has been a wake up call to just how many advantages AI has over humans. The few times I disengaged it because I thought it was going into the wrong lane, it turned out that the car was right and I was wrong. I realized that there is no hope of me driving better than a neural net that knows every road, sees in every direction at once, and never gets tired or distracted. Given that AI has certain inherent advantages over human intelligence, what kind of moats will remain for us as humans? It's a big question. One short-term answer is that the world we live in was created for humans, and in some domains, AI has not closed the gap yet. For instance, AI still struggles to use internet user interfaces. While any computer-literate human can navigate a web page with ease, AI is still not great at making accurate clicks and drags because image embeddings are not optimized for such precision. If the internet were designed to be fed into language models instead of rendered as visual interfaces for humans, AI would obviously far exceed humans. But for now, language models still need to be retrofitted to our legacy infrastructure. Robotics is another area where we humans have a home-field advantage. Most tasks in the physical world are designed around fingers and opposable thumbs, which have been pretty hard to build into robots so far. While it is clear that machines can outperform humans in environments optimized for automation, like large-scale manufacturing lines, for now, most of the world is still built for humans. However, these capability gaps are only temporary. There will surely be a day when machines click faster than us and have superhuman general dexterity. What are the real moats that humans will have? Anything involving private knowledge that language models do not have access to feels like a solid moat to me. Romantic matchmaking and high-end real estate are two examples where inventory is often not advertised publicly and matches are made through being in the right circles. Venture capital is another example—although some research and decision making can be automated with AI, much of success hinges on understanding trends ahead of time and connecting the right people, both of which require private knowledge. While machines can and probably will have increasing access to some types of private knowledge, I think there will still be some types of private knowledge that only humans know. I do not see a path for AI to win when critical knowledge is closely guarded in human circles. A second area where humans seem to have a real moat is in entertainment and the arts, which are inherently valued for their human aspects regardless of how well machines can do them. Watching Usain Bolt sprint one-hundred meters is beautiful as an expression of the peak of human ability, even though cars can drive much faster. Watching chess at the amateur or intermediate level is more relatable and satisfying than watching two superhuman AIs play each other. The value of art comes from the creation process, which is why replicas are not as valuable as originals. These types of work feel like they will continue to have markets even as we advance towards superintelligence. More broadly, human presence is a feature that will be, by definition, challenging for AI to automate. For example, a teacher remembering your name or a parent supporting you is valuable even though AI can easily remember your name and probably give better life advice. Someone spending part of a finite life on you counts because their time runs out. As a personal anecdote, I remember the first time I worked with someone who I considered an amazing AI researcher. His advice was solid but what was more important was that I believed I could do great work with him as a collaborator and I raised my own standards. Over the past decades, the development of technology has divided us in some ways, but hopefully AI brings us closer to a world where human presence is reemphasized. Intelligence has been the defining feature of humans and it will be a big change for AI to automate that over the coming decades. In the near term, certain types of intelligence will become very cheap and automate away old jobs, but the moats I described above will not be the only places where humans can hold value. In the same way that computers took away the jobs of secretaries and manual accountants but created far more jobs via the IT industry, I believe there will be much more demand for services created by productive AI-augmented humans, perhaps for services we cannot yet imagine in today’s society. Just seeing how this story plays out will be an adventure in its own right.