What leading AI figures posted on X
MiniMax M3.1 Flash is now on @tenstorrent Cloud -bringing frontier AI to Tensix silicon, Ethernet scale-out, and an open-source stack from the team led by Jim Keller. Open weights. Diverse architectures. An open, heterogeneous AI future.🫡 [Quoting @tenstorre…
Thrilled to welcome Jeff and the Chroma team to Replit! They’re pioneers in AI. They were building RAG before it was even a thing! At Replit, they’ll help solve one of the hardest problems in AI: giving agents the context and memory they need to do real work.…
Last month a large team of agents used over 100 billion tokens to resolve the Navier-Stokes Millennium Prize problem. Now we're using a similar approach for red-teaming to find and fix infra vulnerabilities. [Quoting @Eric_Wallace_]: To make sure our frontier…
A post on how we've been securing our frontier model training environment. We've been building tools around our frontier cyber models, which have secured us far beyond what we could have unassisted and should generalize as key to how defenders secure themselv…
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of t…
*taps sign* read your own emails your apps send out. particularly if they are being vibecoded. real humans still care and do read them. [Quoting @JohnPhamous]: going viral email before // after
With the explosion of Microduck and Reachy Mini, we'll need more people to join the @pollenrobotics @huggingface teams in all functions. If you're excited about democratizing robotics AI with open-source, my DMs are open!
Thanks @FireworksAI_HQ for bringing M3.1 Flash to serverless on day 0! Looking sharp with the rebrand! 🎆 Try M3.1 Flash in serverless preview for your next multi-step agent workflow 🐰 [Quoting @FireworksAI_HQ]: We have another release you don’t want to miss…
Higher score than Fable 5.1 and Astra 6, wild stuff. [Quoting @ArtificialAnlys]: Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once…
Looking at public benchmarks is a total waste of time No lab worth its salt will release a model without checking and optimizing for benchmarks Literally every model is optimized for benchmarks and 3D renders that can go viral on X
Kimi K3.1 Pro ranking #4 on Artificial Analysis’ Intelligence Index across ALL models, including Claude Opus 5.5, Fable, and GPT-6 Astra, is absolutely wild. Open source is so back. 💪
Kimi K3.1 Pro is out! should be the strongest OSS model now. It beats Claude Opus 5.5 on Terminal-Bench 4.0 at ~1/3 the cost!
We're letting admins have a lot more control over how people get to tokenmax internally by being able to set a floor and a ceiling on the Effort level (which correlates to capability and cost). [Quoting @AskPerplexity]: Enterprise admins can now set Computer'…
TLDR on the kimi k3.1 pro benchmarks is that it looks like they figured out their RL, where the initial release was a very impressive base model with lighter post-training. China is very much keeping pace with capability gains.
one day into the muse japan🇯🇵 launch, muse is the #1 app in japan! ahead of muji, which is perhaps my favorite store in the whole world 🙏
Meet Kimi K3.1 Pro, our most capable 3T-class model.
The only official way to buy Microducks and Reachy Minis is on Given their popularity, a lot of scam websites popped up in the last few days so please be careful. Current orders of Reachy Minis will ship before Christmas so that's the best option if you want…
Kimi 3.1 is cheaper, but it's still not very competitive It's priced at $2 / $10 per 1M input/output Cheaper than K3 but not by 5x. DeepSeek still wins the open-source model race
🚀 Introducing Hy-Nomas: An Open Source Free-Form Interactive Video World Model Key Features: * Free-form interaction with text prompts and camera trajectories * Long-horizon generation with world memory * Lightweight LoRA adaptation for specialized domains s…
Excited to announce Replit for Government and welcome Michael Greenwald to the team! Michael brings a wealth of experience, having previously led AWS’s public sector business. Looking forward to working together to transform how governments build and operate…
Top models being used on security tasks, by spend %: GLM 5.3 & Kimi K3 Grok 4.7 is the top closed-weight model
This user gave Perplexity Computer, SSH access to a Raspberry Pi 5 over Tailscale. Overnight, it autonomously turned it into a game console with an arcade UI, toolchains and MCP integration. The console generates GBA, C, Rust and terminal games with automated…
The cost of intelligence is going to go down by 10x Kimi K3.1 is going to kick it off by offering frontier level models at a fraction of the price Chinese open source labs like DeepSeek will continue to get more performant while maintaining prices
Quick update: we’re still going strong at 13 days, 422 miles, and 1.5 million steps! [Quoting @adcock_brett]: F.03 just crossed 9 straight days walking on this treadmill: 348 miles and 1.2 million steps Started as a side project to stress the robot for any fa…
Computer now has more taste in building websites, thanks to @LexnLin’s open-source skill, now integrated inside. We’re giving 5000 credits per user to tinker around. [Quoting @AskPerplexity]: The Taste Skill by @LexnLin, now in Computer's skills marketplace,…
Using a single Nano Banana image and Blender MCP, Gemini 4 Argon built a 3D Gothic cathedral in @Antigravity. It iterates on the geometry, textures, and lighting to create the finished scene.
Prompt engineering is dead. Prompt however you want. Have fun. [Quoting @HackingLZ]: You’re all prompting wrong
Given the current vulnerability apocalypse, is there anyone left who thinks holding assets in cryptocurrency - where mistaken transactions are irreversible - is still a good idea?
The inventor of the term "prompt injection" suspects (as do I) that this may not be as large a problem in new models. This matches our tests suggesting that you can't simply prompt inject a request for higher grades in a paper, anymore. [Quoting @simonw]: @na…
Slightly bearish on SuperIntelligence until we crack robotics Without robotics we are in the endless treadmill of low value digital loops AI writing docs for other AI to read, writing code for other AI to use and so on 🤷♀️ With robotics, we begin to generat…
The A.I. Daily