Been messing with an uncensored version of Qwen3.8 (27B) running locally in LM Studio, and it's genuinely wild how unfiltered it is. No refusals, no "I can't help with that," no lectures you can ask it literally anything and it just answers straight. Made a video where I used it to break down my own business what I actually need to improve, blind spots I've been avoiding, and where I should be focusing next. It gave way more direct/honest feedback than ChatGPT or Claude typically will, since it's not softening anything. If you're using AI for brainstorming or business strategy, an uncensored local model like this is worth trying curious if anyone else here has messed with it. Dropped the full breakdown here if you want to see it in action:

There's now an uncensored version of Qwen3.8 and it will basically answer anything, no refusals, no lectures, no "I can't help with that." It's called Qwen3.8-27B Uncensored. Someone took Alibaba's brand new flagship open-source model and stripped out the built-in refusal filter, while keeping all of its actual intelligence (coding, reasoning, vision, tool use). Why it's worth knowing about: - Answers straight, no moralizing or refusals on normal questions - Still genuinely smart (real flagship-level model underneath) - Runs locally on your own computer, free, private, no rate limits - Great for builders who need a model that won't randomly block legit use cases Link to the model: https://huggingface.co/orcarouter/Qwen3.8-27B-Uncensored-FP8 If you build with AI, this is worth 10 minutes of your time to try.

been thinking about what I put in here next and I don’t want to guess. everyone’s in a different lane right now — some of you want clients, some of you want your own app in the store, some of you just want agents doing your work for you. the content I make changes a lot depending on which one you’re going for, so vote and I’ll build around whatever wins. if you’re somewhere in between, drop a comment and tell me what you’re working on right now, I read all of them.

Got this comment on one of my reels and it's a fair point, so wanted to explain my thinking. Discord is great for real-time chat with people who already know each other. But as a place for someone brand new to actually discover you, stick around, and get value, it falls apart fast. If a server goes quiet for even a few days it feels dead, old messages scroll away and disappear, and there's no real structure to browse past discussions or find what's actually useful. New people join, see a wall of unread messages, and just leave. Skool solves all of that for me. Posts stick around and stay organized, people can actually search and browse old content, and there's a clear structure (classroom, community, etc) instead of just a chat log. It's way easier for someone to land in my community for the first time and immediately find value, instead of needing to "catch up" on a live conversation they missed. Discord works if you already have a highly engaged, always-online crowd. For building something that grows and holds value over time, Skool has just been a much better fit for me.

I just spent 3 hours cross-referencing every major model drop this week here's the actual data, not the hype. Been tracking every new model release the last few days and figured this community deserves the full rundown instead of scattered posts. Here's the state of play: 🧠 Reasoning & Agentic Tasks DeepSeek V4-Pro-0813 is putting up serious numbers on Terminal-Bench 2.1 (87.9) and Cybergym (83.3), beating GLM-5.2, Kimi-K3, and even Opus-4.8 on several agentic benchmarks. Fable 5 (with fallback) still edges ahead on HLE and DSBench-FullStack though. 💸 Price vs Performance Gemini 3.7 Flash is the standout here $0.75/$3.75 per million tokens while matching or beating Claude Sonnet 5 and GPT-5.6 Terra on things like GDM-MRCR (97.0%) and LVBench (85.4%). If you're optimizing for cost-per-quality, this is the one to watch. ⚡ Speed Not even close — Gemini 3.7 Flash (high) hits 340 tokens/sec output, more than 2x the next closest (GPT-5.6 Luna at 150). If your app has latency-sensitive UX, this matters more than raw benchmark scores. My take: for agentic/coding workflows → DeepSeek V4-Pro. For cost-efficient production apps → Gemini 3.7 Flash. For pure reasoning ceiling → still watching Opus-4.8 and Fable 5. Drop your own experiences with these models below — curious what everyone's actually shipping with right now.