Squeezing in two Qwopus finetunes of @Alibaba_Qwen 3.6 27B into a single card and fronting it with a vibe coded metering proxy made by Claude Fable Very capable at following skills (if a little weak on concurrency patterns in @golang). Thanks Jackrong and @KyleHessling1 !

Well worth experimenting with. Kyle set up early access for me with @inletsdev and it performed easily as well as Qwopus in my Go evals

My new favourite theme for Superterm (.dev) Theme: Catppuccin Mocha Font: Berkeley Mono Free for personal use, commercial use is coffee money.

Time to normalise dual agent use: SOTA vs. best available local for VRAM available. Many believe a time is coming when coding plans will fade away and smaller teams won't be able to afford real token costs. Get familiar with strengths/weaknesses and clear gaps for your workflows and products. Prompting is a skill - and so are local models. This is what we've found helps: * More extensive planning and breaking down tasks into smaller chunks * Additional guidance on architecture and implementation * Explicit notes on testing, unit and e2e * Fine-tunes like Qwopus and the uncoming Qwopus-Coder variant * Agent Skills - will turn a 27B into an expert user of your product * AGENTS.md - place in each repo with specific notes and where to find conventions Pictured - both Opus and Qwen 3.6 27B find the same issue, and offer a similar solution. Opus offers broader insights, but we can still use our own wetware to get more out of Qwen 🧠

llama.cpp now has an official website: llama.app Our goal is to make local AI accessible to everyone, and improving the user experience is a big part of that. On the new landing page you’ll find a single-line cross-platform installer. The installation provides a single unified `llama` entrypoint which you can use to run/serve models and interface with 3rd-party agentic applications. While oriented towards simplified user experience, the new `llama` application also provides all the advanced functionality of the existing llama.cpp tooling with which experienced users are already familiar. Also note that all GGUF models that you might have already downloaded with llama.cpp in the past will be automatically available to use without downloading again (they are stored in the common HF cache on your machine). We have many improvements in the pipeline both at the UX and at the engine level and we plan to iteratively ship new things over the coming months. One of the main focuses will be seamless integration with local-friendly 3rd-party agents (such as Pi). In the meantime, we’ll continue to listen for feedback from the community and adjust accordingly, so keep letting us know what you think and need.

My first experience of an HTTP_PROXY was when I bypassed one at school, and was suddenly everyone's friend! Here's how made them more usable for @slicervm and AI agents slicervm.com/blog/look-ma-no…

Kudos to @alexellisuk for making superterm.dev compatible with SSO using #authentik through OIDC. Works brilliant now I can simply logon using my own credentials instead of the generated token on startup.

Superterm.dev now supports Single Sign-On: GitHub (Device flow - built-in) GitHub (BYO client_id/client_secret) OIDC (Keycloak, Dex, Okta, etc) "superterm update" to get it then open command pallet and "Login & Authentication" for the explainer.

BREAKING! Qwopus 3.6 27B is LIVE! Thank you for your patience on this one, but I believe you'll find the wait was worth it! We've benchmarked this thing up and down, verified that it holds at least a 75.25% (152/202) in the initial 202 SWE bench solves. Not a full run of 500, but it shows the agentic coding quality from the original 27B is retained while adding all of the additional Qwopus benefits across many domains. As always, Jackrong is absolutely cooking here! COT quality has improved significantly through the inversion techniques from our Negentropy proof of concept. It also went through thorough curriculum training. You can check out the MMLU pro benchmarks on the model card, but it improved a whopping 10 points over the base model in physics, as well as meaningful jumps in Chemistry, business, and computer science. However, the best part is that I was able to build an entire survival shooter game using this local model entirely. I genuinely was blown away by the results, which you can play right now on my HF space (link in comments below). "Qwopus Commander" was completed in 9 turns of Qwopus 3.6! To test the new long context training, I made it re-output the entire 3000 line program each turn, and it would make fixes and add features that I requested in large prompts, while perfectly replicating the entire rest of the game from context. What's more is that I did it all at Q8 KV cache quantization, and never had an issue over the entire 303k token run! IMPORTANT: Run it at --temp 0.75 to 1. Mess with it in that range for your use case. Higher temp actually lets the fine-tune shine and be exploratory and is also more stable. Swe Bench was run at temp 1, the game was built mostly at 0.8! We're so blessed to have all of you here and using the models! The support means so much! Please let me know what you build with it in the comments! Or if you have any issues getting it up and running, I will try my best to get back to you! Looking forward to seeing what you legends produce with it this weekend! huggingface.co/Jackrong/Qwop…

Coffee break with codex

Really nice addition to @slicervm, the proxy now supports GitHub Apps. Spin up a fresh sandbox and the proxy uses a GitHub App to inject credentials on the fly. Any action the app has permissions for can be performed without any secrets touching the VM.