Qwen3.5-9B — GGUF Q4_K_M, for Twoody
Qwen/Qwen3.5-9B, converted to GGUF and quantised to Q4_K_M by Osmove for Twoody, the free AI that runs on your own computer. It is the file Twoody installs in one click, and it runs with any app built on llama.cpp.
| File | Size | SHA-256 |
|---|---|---|
Qwen3.5-9B-Q4_K_M.gguf |
5780090688 bytes | 8c4cc24f54473f3d4e080d4ec7fd2a40d0bbec0f94f3c807cd62694905357d54 |
How it was made
- Weights: Qwen/Qwen3.5-9B at revision
c202236235762e1c871ad0ccb60c8ee5ba337b9a, the publisher's own. - Tools: llama.cpp
b11146, the release Twoody ships. The weights were converted to BF16 (convert_hf_to_gguf.py --remote Qwen/Qwen3.5-9B --outtype bf16), then quantised (llama-quantize … Q4_K_M), with no importance matrix. The converter's defaults were kept. - Anyone can make it again with the same weights and the same llama.cpp release, and compare the SHA-256.
Nothing was changed in the model beyond this conversion: its behaviour, its limits and its licence are the original's.
Licence
apache-2.0, as the original model — see LICENSE, copied from
Qwen/Qwen3.5-9B.
Vision
mmproj-F16.gguf (918165888 bytes, SHA-256 2f7ce3652232b91c2220fad23f5b62d110f67e84bd5ed33a7d2bea82f8ee1d26) is the model's vision part, so that
it reads images: converted by Osmove from Qwen/Qwen3.5-9B
at revision c202236235762e1c871ad0ccb60c8ee5ba337b9a with llama.cpp b11146 (convert_hf_to_gguf.py --remote Qwen/Qwen3.5-9B --mmproj --outtype f16).
Give it to llama.cpp with --mmproj.