# Open-source panel lets users run llama.cpp, stable-diffusion.cpp and audio.cpp locally

Digest AI · Agents & Tools · published 2026-09-23T20:54:13Z

Canonical: https://digestai.news/story/open-source-panel-lets-users-run-llama-cpp-stable-diffusion-cpp-and-au

## Summary

A community‑maintained control panel provides a single‑page web UI for running local AI inference on a personal GPU box. It supports llama.cpp for text and vision chat, stable-diffusion.cpp for image generation, and audio.cpp for text‑to‑speech, voice cloning, transcription, music and sound effects. The panel is a stdlib‑only Python server with zero pip dependencies, systemd‑user units, and no Docker or database requirement.

The UI lets users create any number of inference instances, each pinned to a chosen GPU or CPU, with independent parameters, ports and logs. Over 200 settings are grouped with plain‑English tooltips, and a VRAM/RAM estimator refuses launches that would exceed available memory. Features include fallback tiers that downgrade configurations after repeated failures, an optimizer that benchmarks settings against coding and agent workloads, and a decode‑vs‑depth curve that measures tokens per second at various context lengths while tracking GPU temperature. Models can be downloaded directly from Hugging Face, and the panel offers an in‑browser terminal, file manager, and detailed crash forensics.

The panel runs on Ubuntu 26.04 with Python 3.14, Caddy 2.x for TLS and basic auth, and ttyd for the terminal. Reference hardware in the documentation is an AMD Radeon 7900 XTX, but the setup also supports NVIDIA via nvidia‑smi and Vulkan‑NVK. All configuration is stored in plain files, making backup and migration straightforward.

## Key points

- Panel runs llama.cpp, stable-diffusion.cpp and audio.cpp from one browser tab without Docker or a database.
- Supports multiple GPU‑pinned instances, 200+ settings, VRAM/RAM estimator, fallback tiers and optimizer benchmarks.
- Built with stdlib‑only Python, systemd units, Caddy TLS and works on Ubuntu 26.04; reference hardware is a Radeon 7900 XTX.

## Why it matters

It lowers the technical barrier for hobbyists and developers to host large language, image and audio models on personal hardware, reducing dependence on cloud services.

## Sources

1. [Self-hosted control panel for local AI on your own GPU box: run llama.cpp chat models, stable-diffusion.cpp images and audio.cpp speech, music and sound from one browser tab. Instances per GPU, 200+ explained settings, launch previews, VRAM/RAM estimates, benchmarks, build updates, power caps & CLI](https://github.com/W61k3r/LexiPanel) (github.com, 2026-09-23, primary source)

Part of the developing story: [Rise of Local Open-Source AI Tools](https://digestai.news/thread/kdnuggets-lists-seven-open-source-chatgpt-alternatives-that-run-locally) (2 stories)

## Cite

Digest AI, "Open-source panel lets users run llama.cpp, stable-diffusion.cpp and audio.cpp locally", 23 September 2026, https://digestai.news/story/open-source-panel-lets-users-run-llama-cpp-stable-diffusion-cpp-and-au

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/open-source-panel-lets-users-run-llama-cpp-stable-diffusion-cpp-and-au.json