Local AI harness/workstation: local AI image editor, Ollama image generation routing, SDXL inpainting UI, CivitAI model imports, 8GB VRAM Stable Diffusion.
-
Updated
Nov 13, 2026 - Python
Local AI harness/workstation: local AI image editor, Ollama image generation routing, SDXL inpainting UI, CivitAI model imports, 8GB VRAM Stable Diffusion.
Local LLM + voice-cloning TTS sharing one 8 GB GPU, as a server for ESP32 voice boards. 本地 LLM + 声音克隆 TTS 共用一块 8 GB 显卡,给 ESP32 语音板子当服务端。怎么省显存、怎么轮流、怎么不断音,附完整实测。
Archive of every local LLM tested on one NVIDIA RTX 2080 8GB (TU104) on Linux — recipes, runtimes, quants, pinned sources, settings, context ceilings, and reproducible measurements. Failures included.
Run a coding agent on an 8 GB consumer GPU. 4 models optimized for best performance, with a launcher for any harness.
Running Stable Diffusion Forge with uncensored community models on an 8GB VRAM laptop via WSL2.
Deploying personal-grade production models on a consumer-grade PC
Run a 35B LLM on a GTX 1070 with 8 GB VRAM: custom CUDA inference engine for Ornith-1.5-35B-A3B (MoE, NVFP4) on Windows, ~25-30 tok/s. Expert streaming, fused NVFP4 kernels, prefix cache, OpenAI-, Anthropic- and Ollama-compatible API for local AI agents.
Everything I made run fast on a 2016 GPU with 8 GB: llama.cpp kernels, CUDA graphs, speculative decoding, ternary models, LoRA training, video diffusion on a GTX 1080 (sm_61). Measured, with sources.
ComfyUI sidebar panel + 3 nodes for the H3 Director's Desk video pipeline, with a deployable service copy.
Poster8G — run a poster print shop on your own 8 GB GPU. Local AI background, real layout, cutout, QR, print-ready PDF. For DeepSeek Harness.
Drive a LOCAL ComfyUI from Claude Code / Cursor / any MCP harness - tuned for low-VRAM (8GB) rigs and heavy video models (MiniMax H3). Paste-ready MCP config (artokun/comfyui-mcp) + a pure-REST fallback skill. MIT.
8GB 显存笔记本上搭 AI 角色:SDXL 出图 25 秒、GPT-SoVITS 语音 1.5 秒、ControlNet 姿势控制。全程本地推理,零 API 费用。附 26 个踩坑记录。
To associate your repository with the 8gb-vram topic, visit your repo's landing page and select "manage topics."