List of supported models (and sticky)?

[brunette69-ruby]

Where can I find list of supported model?
Embedding: ?
Rerank: ?
Generate: ?

Thx

[ExtReMLapin]

type gguf in huggingface search bar

[brunette69-ruby]

Not quite. Some seem to be, some not. Not format wise (gguf is there), but function wise - tokenization doesn't work.
For example:
Qwen3 are widely available in GGUF and often best in class (embedding and reranking). But not supported.

ggerganov
on Jul 6, 2025

Qwen3 Embedding and Qwen3 Reranking models are not supported yet.

on Jul 6, 2025

Also the GGUFs provided by Qwen are missing important metadata and will not work until fixed and supported.

[ryan-mangeno]

https://github.com/ggml-org/llama.cpp/blob/master/src/llama-arch.h will show you supported architectures, mtmd is being worked towards so vlms wont be immeditaly obvious, finding embeddings ( or tensor names ) you can use the llama-eval-callback script to dump the tensors and to generate gguf from hf you would use the convert_hf_to_gguf to generate .gguf files from their hf impl