Reddit user says ChatGPT Plus ban over 'cyber abuse' was reversed

A LocalLLaMA user reports that Strata run configs lacking a `sampling` block decoded greedy on builds before 0.1.45 whenever the request set no sampling of its own. The setup ran on an RTX 4090, i9-14900K and 64GB DDR5-5800, moving from IQ2_XS to Unsloth's UD-IQ4_XS.

Free GIMP 3 plugin routes ComfyUI workflows into Adobe-style Generative Fill, Remove and Expand tools without a node graph. Requires ComfyUI and a decent GPU.

Post collects Claude Code workflows from blogs, X, and GitHub, recommending installing 2-3 subagents from VoltAgent's awesome-claude-code-subagents repo, which lists 160+ definitions.

The manifesto argues 'AI' is not a neutral tool but one built to control populations and substitute human judgment, and calls for its abolition as a political project. It draws parallels to slavery, prison, police, and family abolition, framing abolition as working toward a liberatory alternative rather than mere opposition.

The tiny web app for browsing and editing files in a folder, built mainly for editing files on different machines via Tailscale, adds a copy-to-clipboard option for both text and image data. Willison wanted an easy way to move pelican prompt logs into a Gist from his phone.

A Reddit user says Claude Opus 5.5 produced a credible first-pass design for an open-source MRI scanner costing roughly $31k in a few hours. The author stresses it is simulation-only: nothing has been built or tested, so it may not work.

A LocalLLaMA poster reports running Qwen 3.8 27B at 262K context using an Unsloth IQ4_XS quant with KV cache streaming, arguing pricey multi-GPU builds aren't required for local LLMs.