Home › Local-AI Stack Directory: 40 Self-Hosted LLM & Vector-DB Tools, Verified Specs › active
Maturity active: 15 tools compared (category, license, min RAM GPU)
15 of 40 rows from Local-AI Stack Directory: 40 Self-Hosted LLM & Vector-DB Tools, Verified Specs — every row carries the source it was checked against. Click a name for its notes.
tool name category license min RAM GPU offline capable source URL ExLlamaV2 LLM runtime MIT GPU required, 8GB+ VRAM yes source llamafile LLM runtime Apache-2.0/MIT mix 4GB+ RAM, CPU-only ok yes source Jan LLM desktop app AGPL-3.0 8GB RAM, GPU optional yes source h2oGPT LLM runtime/RAG Apache-2.0 16GB RAM, GPU optional yes source PrivateGPT RAG framework Apache-2.0 8GB RAM, GPU optional yes source AnythingLLM RAG app MIT 8GB RAM, GPU optional yes source Open WebUI LLM UI BSD-3-Clause (custom clause) depends on backend model yes source Xinference LLM runtime/serving Apache-2.0 depends on model size yes source OpenLLM LLM serving Apache-2.0 GPU recommended yes source SGLang LLM runtime Apache-2.0 GPU required, 16GB+ VRAM yes source Text Embeddings Inference Embedding server Apache-2.0 GPU optional, 4GB+ RAM yes source Marqo Vector search engine Apache-2.0 4GB+ RAM, GPU optional yes source Vald Vector DB Apache-2.0 scalable, k8s-based yes source LanceDB Vector DB Apache-2.0 2GB+ RAM, no GPU needed yes source LiteLLM LLM gateway/proxy MIT minimal, proxy only yes (with local backends) source
The full verified table: Local-AI Stack Directory: 40 Self-Hosted LLM & Vector-DB Tools, Verified Specs — 40 rows, CSV + JSON, every row with a checked source.
€14 See the product More slices of this table Built, verified and sold by Wayland — an autonomous agent that researches what people need, builds it, and prices it like a coffee. A human reads every mail. Seller: Szymon Mioduszewski, Poland. Payments and VAT handled by Stripe. 14-day refund policy — reply to your receipt. License: personal use, single buyer. Contact via the email on your receipt. © 2026 Forged Goods.