DEDUPED Hot AI repos, turned into use cases Get 5 a day

What you can do with Kyojin

Kyojin is an open source inference engine, built on ExLlamaV3, that runs large GLM, Qwen and MiMo models on one AMD Strix Halo mini PC with 128 GB of memory.

Trended earlier

Yamz-Labs/kyojin on GitHub

Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk.

Authorization notice. This tool is dual-use. Only use it on software, sites and systems you own or have explicit permission to test, and follow the relevant licenses and terms of service.

What do you want to do?

5 use cases

  • Run a local OpenAI-style LLM server on AMD

    Serve a 95 GB Qwen model on one 128 GB AMD mini PC through an OpenAI-compatible API on port 8000.

    View prompt
    In the Yamz-Labs/kyojin repo, follow doc/install.md, then download the Qwen3.8-Flash-Next pack and start tools/qwen/serve.py on port 8000 on <target host>. Expected output: the server starts and answers one OpenAI API chat request with text. Verify that the response JSON includes a choices array before reporting success.

  • Read a 500K-token document in one prompt

    Load a very long report or codebase dump and query it locally with context up to 500K tokens.

    View prompt
    In the Yamz-Labs/kyojin repo, start tools/qwen/serve.py with -c 524288 and --cache-bits 8 for the Qwen pack, then send <target document> as a single prompt. Expected output: an answer that cites facts from the document. Verify by checking two facts from the start and end of the text against the source before reporting.

  • Benchmark a Strix Halo machine and share results

    Measure prefill and speculative decode speed on your own 128 GB AMD box and export the numbers as JSON.

    View prompt
    In the Yamz-Labs/kyojin repo, run tools/bench.sh --json on <target machine> after reading doc/benchmarks.md. Expected output: a JSON benchmark result with prefill and speculative decode tokens per second. Verify that the figures are labeled with the context length and model used before comparing them to the README tables.

  • Scrape server metrics from a local model host

    Expose Kyojin's Prometheus metrics endpoint so a monitoring stack can track a local model server.

    View prompt
    In the Yamz-Labs/kyojin repo, start the Qwen server on port 8000 and confirm the Prometheus /metrics endpoint responds on <target host>. Expected output: a plain text metrics page from the server. Verify that the response contains metric lines before adding it as a scrape target.

  • Load a less-restricted model variant locally

    Run an optional uncensored mode on GLM, MiMo or the Qwen preset for controlled red-team and research tests.

    View prompt
    In the Yamz-Labs/kyojin repo, read doc/refusal_hook.md and load the uncensored variant of <target model> following doc/models.md. Expected output: the server starts with the mode applied and returns responses to a scoped test prompt. Verify that the test is covered by an approved plan and that all requests are logged before running it.

How to use Kyojin

  1. Check you have what it needs: AMD Ryzen AI Max+ 395 mini PC with 128 GB of memory (Strix Halo, gfx1151); ROCm build of torch, installed per doc/install.md; Python and the hf command-line tool to download model packs.
  2. Open Yamz-Labs/kyojin on GitHub and read the README and license.
  3. Install ROCm torch and build Kyojin once by following doc/install.md.
  4. git clone https://github.com/Yamz-Labs/kyojin && cd kyojin
  5. hf download yamz-labs/Qwen3.8-Flash-Next-EXL3-Yamz --local-dir ./qwen-pack
  6. python tools/qwen/serve.py --model ./qwen-pack --port 8000 -c 131072

These steps come from our top use case, Run a local OpenAI-style LLM server on AMD, which also has a copy-paste agent prompt.

Kyojin alternatives · Best AI repos for Software and DevOps

About this repo

  • Who it is forWritten for developers and self-hosters who own an AMD Ryzen AI Max+ 395 machine with 128 GB and want a local OpenAI-compatible model server.
  • Why it is hot
  • LanguagePython
  • Sourcegithub.com/Yamz-Labs/kyojin · Back to Hot repos

Requirements

  • AMD Ryzen AI Max+ 395 mini PC with 128 GB of memory (Strix Halo, gfx1151)
  • ROCm build of torch, installed per doc/install.md
  • Python and the hf command-line tool to download model packs
  • Roughly 100 GB of free disk space for a model pack (Qwen pack is 95 GB)
Maintain Kyojin? Add the Featured on Deduped badge

Paste this into your README:

[![Featured on Deduped](https://dedupedai.com/badge/yamz-labs-kyojin.svg)](https://dedupedai.com/use-cases/yamz-labs-kyojin/?ref=badge)

More badge styles and HTML

Questions about Kyojin

What is Kyojin?

Kyojin is an open source inference engine, built on ExLlamaV3, that runs large GLM, Qwen and MiMo models on one AMD Strix Halo mini PC with 128 GB of memory. It is an open-source project on GitHub: https://github.com/Yamz-Labs/kyojin.

Who is Kyojin for?

Written for developers and self-hosters who own an AMD Ryzen AI Max+ 395 machine with 128 GB and want a local OpenAI-compatible model server.

What can I do with Kyojin?

5 use cases so far, for example: Run a local OpenAI-style LLM server on AMD; Read a 500K-token document in one prompt; Benchmark a Strix Halo machine and share results. Each has steps and a copy-paste agent prompt.

What do I need to run Kyojin?

AMD Ryzen AI Max+ 395 mini PC with 128 GB of memory (Strix Halo, gfx1151); ROCm build of torch, installed per doc/install.md; Python and the hf command-line tool to download model packs; Roughly 100 GB of free disk space for a model pack (Qwen pack is 95 GB).

What are alternatives to Kyojin?

Other hot AI repos in Software and DevOps: Harness2, iFixAi, Orca, Diagram Design, AI Memory.

Is Kyojin safe to run?

Deduped does not own, audit or vouch for it. Read the README and license, run it on your own systems first and review every command before you run it.