Thireus' GGUF Recipe Maker · Simple mode
Produce optimum GGUF recipes calibrated to your hardware with the GGUF-Tool-Suite! Use the toggle at the top-right to switch to the expert UI.
Quantization selection preset (choose either ik_llama.cpp or llama.cpp quants)

Thireus' GGUF Recipe Maker · Expert mode

A web-based port of quant_assign.py and quants_regex_merger.py • Automated GGUF .recipe file production
✔ Firefox / Chrome / Safari ready
v2.7.0
GGUF-Tool-Suite  ·  Thireus' Model Collection  ·  GGUF Quant Downloader  ·  💖 Donate  ·  Model recipes are produced using the GGUF Tool Suite.
Step 1: Select a preset, model and a target size  ·  Step 2: hit the "Load & Patch Scripts" button  ·  Step 3: hit the "Run" button  ·  Step 4: upload the .recipe obtained to GGUF Quant Downloader

Presets, host & model inputs

Global settings
New to the GGUF Tool Suite? Try the Qwen3-4B models. They are small and ideal for initial testing.
Huggingface Checking…
gguf.thireus.com Checking…
GitHub Checking…
bf16 model size: N/A

Simple parameters

Runner settings
➤ Display perplexity prediction graph
50.00%
2.14 GB (1.99 GiB) - 4.25 bpw (~ 11.06 ppl) ~ iq4_ks
Status: Idle
No dataset files loaded.
No recipe available yet

Advanced parameters

Cmd & recipe logs