Radiance-LLM

Radiance is a modular inference engine for AMD GPUs that serves language models over OpenAI chat/completions/Responses, Anthropic Messages, and llama-server…

  • AI
Radiance is a modular inference engine for AMD GPUs that serves language models over OpenAI chat/completions/Responses, Anthropic Messages, and llama-server APIs so existing clients work unchanged, and also generates images, video, speech, and music from a single server with a live dashboard. It targets RDNA 3, RDNA 4, and Ryzen AI Max cards (RX 9000 and RX 7000 series, Strix Halo) and needs no ROCm, compiler, or driver plugin on the host. This Unraid template runs the official stilldeadcode/radiance image (Apache-2.0) and suits home-lab users who want one AMD GPU machine for both LLM chat and media generation.

Platform availability

Available on 1 of 10 platforms

  • Unraid
  • TrueNAS (not listed)
  • Umbrel (not listed)
  • ZimaOS (not listed)
  • Proxmox (not listed)
  • Helm (not listed)
  • Mac (not listed)
  • Windows (not listed)
  • Linux (not listed)
  • Any server (not listed)

Health score?A 0-100 score computed nightly from four factors: maintained (40%: recent commits and steady activity), popular (25%: GitHub stars, log scale), easy to install (20%: platforms, install notes, setup guides, Docker image) and light to run (15%: minimum RAM, ARM support). When the source code has been checked, a fifth factor, security check (20%: automated scan for committed secrets, unsafe desktop-app settings, known-vulnerable dependencies and project hygiene; any serious finding caps it at 20), is added. Factors without data are left out and the rest are rescaled.

59/100 Fair

Maintained
100last commit 11 days ago
Popular
45167 GitHub stars
Easy to install
351 install platform, install notes, Docker image
Light to run
No data
Security check
201 serious finding, 2 minor, CI, tests

Based on 4 of 5 factors.

Availability: platform listings confirmed today

GitHub stars
167
Open issues
5
Last commit
2026-11-10
Activity
steady

Checked 10 days ago - source: GitHub (magiccodingman/vllm-radiance)

Resources & Compatibility

ARM (e.g. Raspberry Pi) is not supported.

Checked 10 days ago - source: Docker Hub image tags and published docs

First-install notes

  • Unraid model placement and firmware Put model files on a fast local pool (ext4, XFS, or btrfs, e.g. /mnt/cache), never on /mnt/user, because the engine reads them via direct I/O. Enable Resizable BAR (Above 4G Decoding) in the firmware. source
  • Supported GPUs and host dependencies Radiance supports RDNA 3, RDNA 4, and Ryzen AI Max cards (RX 9000 and RX 7000 series, Strix Halo) and does not require ROCm, a compiler, or a host driver plugin. source

Fetched 5 days ago - each note links to its source

Alternatives

No alternatives collected yet.

Common questions

Does the host need ROCm installed to run Radiance-LLM?

No. It needs the amdgpu kernel driver, which Unraid ships on Linux 6.8+, and the host needs no ROCm, compiler, or driver plugin.

What firmware setting should be enabled before using an AMD GPU?

Resizable BAR, also called Above 4G Decoding, should be enabled in the firmware.

Where should model files be stored?

Model files should be on a fast local pool, with ext4 or XFS indicated.

Can existing OpenAI, Anthropic, or llama-server clients connect without changes?

Yes. It serves OpenAI chat, completions, and Responses APIs, Anthropic Messages, and llama-server APIs so their clients work unchanged.

Where is the built-in dashboard available?

A live dashboard is served at the root URL.

Answers sourced from codeberg.org

Community

No community discussions collected yet.