LLM-Server

Files

T

haylanandClaude-Bot ac3f730f83 docs(research): recommend Qwen-Image FP8 for full-VRAM diffusion build

Closes issue #39 — with llama-server stopped and the full ~32GB R9700
available, Qwen-Image (Apache-2.0, 20B MMDiT) at FP8 precision (~25GB)
is the recommended upgrade from FLUX.1-schnell: it has the cleanest
license of the candidates and is the only one with a ComfyUI workflow
pre-validated specifically on this GPU architecture (gfx1201/R9700),
per kyuz0/amd-r9700-comfy. HunyuanImage-3.0 is ruled out (CUDA-only,
multi-GPU datacenter VRAM floor); Krea-2 flagged as promising but too
new for R9700-specific field evidence.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015zCwaWJQuKgXDUfRPBqDS7

2026-09-05 21:03:13 +02:00

connecting-cloud-providers.md

research: Claude Pro subscription vs Anthropic API for gateway backends

2026-08-26 08:51:38 +02:00

image-generation-model-choice.md

docs(research): recommend Qwen-Image FP8 for full-VRAM diffusion build

2026-09-05 21:03:13 +02:00

langchain-pgvector-vs-litellm-pgvector.md

docs(research): LangChain+pgvector-direct vs litellm-pgvector for issue #25

2026-09-02 22:02:58 +02:00

lazytainer-omniroute-idle-stop.md

docs(research): confirm lazytainer/omniroute idle-stop conflict (#40 )