Alright I finally have more to share on my personal project! Resin is a runtime for building fast LLM inference engines. The idea is to make all the generic parts of inference reusable, and leave only the model-specific forward pass + kernels to be written for each architecture / hardware target. To build an engine: 1. Import Resin's core. 2. Have an agent write a forward pass for the model you’re interested in, for your specific hardware (GPU or TPU or AMD), any kernels you want. Then, boom, you have a standalone inference engine. — I've been hacking over the past month (early post: xitter.cf/ekzhang1/status/208950…), and recently this week I got a lot of features + parallelisms working, inspired by DeepInfer. For DeepSeek-V4.1-Flash, it's nearly as fast as DeepInfer (within 5-10%), while being self-contained and supporting other models, plus metrics + HTTP server + prefix cache. Yesterday I wrote a skill for authoring models and put Codex to work on it overnight. The agent built a GLM-5.3-Flash engine with MTP, reaching up to 60,000 input tok/s and 16,000 output tok/s on B200 (EP4). So far inference "vibe coding" has not been very reliable or successful. I think it's because people start from scratch each time and reinvent lots of tricky things on the scheduler and server layers. But with a shared core, we can automate lots of inference work. The tricky bit is writing a fast forward pass. But Resin turns that into a standalone engine by just importing a library, and it becomes a verifiable goal for the AI. Code: github.com/ekzhang/resin

Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model. 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal. 🔹 Full-project coding: games in Unreal, Unity & Godot. 🔹 Excels at parametric 3D modeling in Blender & CAD. 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%. 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks. Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API. Open weights by Nov 4, 2026. 🔗 API: platform.kimi.ai 🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro

Android phones deserve bottom sheets too ❤️

We formally announced Mirendil less than 4 months ago. Today, I'm excited to share an important milestone for: Pretraining at scale without a pretraining team. Mirendil's mission is to solve Recursive Self-Improvement (RSI) and use it to democratize frontier AI development, ultimately accelerating science and technology. At Mirendil, when we think about RSI, we think holistically about the company as an intelligent entity that self-improves toward its goal. From that perspective, the most powerful notion of acceleration is reducing the time and resources (number of human experts, compute, and other costs) needed to achieve ambitious goals. Given that pretraining is perhaps the most important yet most challenging part of frontier AI development, we decided to run an ambitious controlled experiment: Starting fully from scratch, can we significantly reduce the time, number of human experts, and compute resources needed to go all the way to successfully pretraining a strong coding model at scale? Mirendil-Base is the result of this experiment. Our first pretrained model was built by one researcher in 3.5 months, with less than $10M in compute. This includes all the research and small-scale experiments it took to get there, as well as the final training run. The recipe holds up as the models get bigger. For reference, Mirendil-Base is in the 100B-500B parameter range, Mirendil-Base-Large in the 500B-2T range, and Mirendil-Base-XL in the 2T-10T range. We're beginning to see a world where a single researcher, empowered by AI, can take on work that once required an entire research organization. As these systems improve, the bottleneck shifts from the size of a team to the ambition and judgment of the people directing them. This has profound implications for who gets to build AI and, ultimately, who gets to push the frontiers of science and technology forward. We’ll share independent evaluations, post-training benchmarks and API-access sooner than you think so stay tuned!

New research: We developed JOLT, a post-training method where one model acts as both teacher and student, with the teacher trained to give feedback the student can use. On MATH, it matched GRPO's accuracy with 6.5x fewer training tokens. perplexity.ai/hub/blog/a-goo…

New on the blog: — You could do it alone Solo company is a new idea, but it think it's bit like the TV show Alone: Every decision, task and outcome is yours, day after day, and there is no one else to attend the fire karrisaarinen.com/posts/you-…

An urgent notice from the Railway team blog.railway.com/p/security-…

AI has read the whole internet. But it’s never run payroll or a business. We’re changing that. Today, we’re launching Emet, our AI lab for model evaluation and training, and Pangaea, our first family of models for global work. On real payroll tasks, the best AI model fully solves only 49%. Close enough to look right. Wrong enough to break a paycheck. At Emet, in-country experts define what getting it right looks like. We turn that into public benchmarks and training data, and score every model against them. Ours included. Pangaea is the first proof. Built on open-weight models and trained with Emet, it gets close to top-model accuracy, up to 3x faster, at a fraction of the cost. @deel and the payroll teams that joined us bring decades of experience running payroll, HR, and compliance across 150+ countries. Knowledge the internet doesn’t have. Emet puts it to work for every AI model, not just ours.