Consultant — Confidential Documents — RTX 4070
AnythingLLM + Ollama + Gemma 3 12B IT (QAT/int4 profile)
GOODfully-local
Directory
Reference stacks
Evidence-backed configurations used by the launch Finder.
Consultant — Private RAG — Apple 32 GB
AnythingLLM + Ollama + Qwen3-8B
GOODfully-local
Legal — Confidential Documents — Apple 64 GB
PrivateGPT + llama.cpp + Mistral Small 3.1 24B
EXCELLENTfully-local
Legal — Private RAG — RTX 4090
Open WebUI + Ollama + Mistral Small 3.1 24B
EXCELLENTself-hosted
Developer — Agentic Coding — Apple 64 GB
OpenHands + LM Studio local server + Qwen3.6-35B-A3B
GOOD_QUANTIZEDlocal-network
Developer — Agentic Coding — RTX 4090
OpenHands + LM Studio local server + Qwen3.6-35B-A3B
GOOD_QUANTIZEDlocal-network
Developer — Local Coding — RTX 4060 8 GB
Continue + Ollama + Qwen2.5-Coder-7B-Instruct Q4_K_M
GOOD_WITH_BOUNDED_CONTEXTfully-local
Finance — Document Analysis — RTX 4070 Ti SUPER
AnythingLLM + Ollama + Qwen3-14B
GOODfully-local
Finance — Private RAG Server — L40S
Onyx + vLLM + Mistral Small 3.1 24B
EXCELLENTself-hosted
Cybersecurity — Security Log Analysis — RTX 4090
Open WebUI + Ollama + Qwen3-14B
EXCELLENTlocal-network
Cybersecurity — Malware Report Analysis — RTX 4070
AnythingLLM + Ollama + Qwen3-8B
EXCELLENTfully-local
Research — Private Research Assistant — Apple 32 GB
Khoj + Ollama + Gemma 3 12B IT (QAT/int4 profile)
GOODfully-local
Consultant — Private Meeting Transcription — RTX 4060
whisper.cpp + whisper.cpp + Whisper large-v3-turbo
GOODfully-local
Creative — Local Image Generation — RTX 4090
ComfyUI + ComfyUI backend + Stable Diffusion XL Base 1.0
EXCELLENTfully-local
Sensitive / Regulated — Air-Gapped Assistant — CPU 32 GB
Open WebUI + llama.cpp + Qwen3-4B
GOOD_BUT_SLOWER_THAN_GPUair-gapped