Views
No views yet
| File | Size | Notes |
|---|---|---|
cyberneurova-Qwen3.8-27B-Q2_K.gguf | ~10 GB | smallest, runs on modest hardware |
cyberneurova-Qwen3.8-27B-Q3_K_M.gguf | ~13 GB | small |
cyberneurova-Qwen3.8-27B-Q4_K_M.gguf | ~17 GB | recommended — best size/quality |
cyberneurova-Qwen3.8-27B-Q5_K_M.gguf | ~19 GB | higher quality |
cyberneurova-Qwen3.8-27B-Q6_K.gguf | ~22 GB | near-lossless |
cyberneurova-Qwen3.8-27B-Q8_0.gguf | ~29 GB | highest-quality quant |
1llama-cli -hf cyberneurova/cyberneurova-Qwen3.8-27B-GGUF:Q4_K_M -p "Hello"
2# or a downloaded file:
3llama-server -m cyberneurova-Qwen3.8-27B-Q4_K_M.gguf -c 8192ollama run hf.co/cyberneurova/cyberneurova-Qwen3.8-27B-GGUF:Q4_K_Mmmproj-F16.gguf (~0.9 GB) alongside any text quant:1llama-server \
2 -m cyberneurova-Qwen3.8-27B-Q4_K_M.gguf \
3 --mmproj mmproj-F16.gguf \
4 -c 8192llama-mtmd-cli, or the
OpenAI-style API with an image_url). One mmproj-F16.gguf works with every
quant in this repo.