The local / self-hosted LLM stack

Running open models yourself for privacy, cost, or control. The stack is a serving runtime, optional fine-tuning, and supporting tooling.

GraphCanon updated 5d · 55 views this month

Inference runtime - serve open-weight models efficiently on your hardware.

When not to use: Self-hosting rarely beats a hosted API on cost until you have steady, high-volume traffic.

Fine-tuning - adapt an open model to your domain with LoRA or full fine-tunes.

When not to use: Try prompting and RAG first; fine-tuning is the answer to style/format, not missing knowledge.

Tooling & gateways - a unified API in front of your local models.

When not to use: Direct calls are fine for a single local model with one consumer.

Common questions

What is the the local / self-hosted llm stack?
Running open models yourself for privacy, cost, or control. The stack is a serving runtime, optional fine-tuning, and supporting tooling.
When should I use the the local / self-hosted llm stack?
Use this stack when your constraints match its layers: Inference & Serving, Model Training, Developer Tools. Each step on the page includes when-not-to-use guidance so you do not over-engineer.
How does GraphCanon pick tools for the local / self-hosted llm stack?
Stack pages are editorial workflows over canonical categories. Representative tools are pulled live from the graph at render time (top adoption in each category), not a fixed marketing list.
Where can I compare tools in this stack?
Follow category hubs and head-to-head compare pages linked from each step. Featured comparisons cover the highest-intent pairs (frameworks, vector DBs, agent runtimes). Start at compare hub.
Is there a machine-readable version of The local / self-hosted LLM stack?
Yes. Append .md to `/stacks/local-llm` or fetch `/md/stacks/local-llm` for a markdown twin with steps, when-not notes, and tool links.

Was this helpful?

Anonymous feedback helps us improve pages and translations.