{"slug": "turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad", "title": "Turnstone? Designing Electronics? It's more likely than you think! (A preview of more mad science coming soon)", "summary": "A developer experimenting with the latest model releases reports that DeepSeek 4.1 Flash 1017 and GLM 5.3 Flash both perform well at driving a 3D-accelerated desktop, with the DeepSeek model used for the most recent demo. The developer's rig runs a stack of RTX 5090s for smaller workloads and 4x6k GPUs for a large reasoning model, alongside TTS, embedding, instruct and vision models, while a separate Proxmox machine hosts 3D desktop VMs on older Ampere and Ada-generation RTX 4K cards via simple PCIe passthrough. The developer also describes building RCON tooling for multi-turn, multi-input tool calling — letting an LLM hold control, inspect the screen, press tab to select options and release control — and patching Turnstone to return multiple multimedia outputs (video/image) from a single tool call, with video embedding tests aimed at click-and-drag manipulation of 3D models.", "body_md": "Deepseek 4.1 Flash 1017 as what i did the latest demo with, GLM 5.3 flash also works well for driving a 3d accelerated desktop. I am always experimenting with the latest releases to see how they do.\n\nThe stack of 5090s are for smaller workloads, the 4x6k’s run the big reasoning model, i also have tts models, embedding models, instruct and special vision models… the rig is a busy place.\n\nProxmox is running on a different machine for the 3d desktop VMs using older ampere and ada generation rtx 4k’s doing simple pcie passthrough.\n\nDeveloping RCON tooling to allow multi-turn multi-input tool calling was a mind bender. How do you let the LLM hold control, and look at the screen, then press tab until it selects the right option, then release control? Clicking and dragging to move 3d models around is where my tests with video embedding come in. Returning multiple multimedia (video/image) from a single tool call is something i had to patch into turnstone.", "url": "https://wpnews.pro/news/turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad", "canonical_source": "https://forum.level1techs.com/t/turnstone-designing-electronics-its-more-likely-than-you-think-a-preview-of-more-mad-science-coming-soon/259031#post_5", "published_at": "2026-10-23 19:49:51+00:00", "updated_at": "2026-10-23 20:11:05.398077+00:00", "lang": "en", "topics": ["ai-agents", "large-language-models", "ai-tools", "computer-vision"], "entities": ["DeepSeek 4.1 Flash 1017", "GLM 5.3 Flash", "Turnstone", "Proxmox", "RTX 5090", "Nvidia"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad", "markdown": "https://wpnews.pro/news/turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad.md", "text": "https://wpnews.pro/news/turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad.txt", "jsonld": "https://wpnews.pro/news/turnstone-designing-electronics-it-s-more-likely-than-you-think-a-preview-of-mad.jsonld"}}