Qwen3.8-27B
UD-Q8_K_XL GGUF + mmproj · unsloth/Qwen3.8-27B-GGUF
done
| Publisher | unsloth |
|---|---|
| Modality | text+vision LLM |
| Parameters | 27.0B 27B dense (model card). Hybrid attention: 48 linear + 16 full layers (config.json). Native context 262,144. Source: model card. |
| Format | UD-Q8_K_XL GGUF + mmproj |
| Size on disk | 30.16 GiB (32,389,149,295 bytes) |
| Estimated memory to run | ~38.0 GB (estimate; see method) |
| Native context | 262,144 tokens (config/card) |
| Engine | llama.cpp / LM Studio / Ollama · recipe |
| License | apache-2.0 Apache 2.0: commercial OK. |
| Archive folder | official |
| Notes | Near-lossless 8-bit. |
Where it runs
| Machine | Verdict | Budget used |
|---|---|---|
| Bosgame M5 (x1) | Fits ≈6.7 tok/s ceiling | ~108-110 GB GPU-usable on Linux with GTT raised (96 GB cap on Windows). Fit budget used here: 108 GB. |
| Bosgame M5 (x2, RPC) | Fits ≈6.7 tok/s ceiling | ~200-225 GB usable across two nodes. Fit budget used here: 210 GB; 210-225 GB = tight. |
| MacBook Pro (M4 Max) | No | macOS caps GPU wired memory by default (~75% = ~27 GB). Raising iogpu.wired_limit_mb to ~32 GB is possible but leaves little for the OS. |
| Mac mini (M4) | No | ~2/3 of 16 GB usable by the GPU by default. Already busy as build box + Hermes/Ollama. |
| omarchy (AMD desktop) | After 128 GB RAM | 7 GB of the 8 GB VRAM for weights. Offload capacity depends on system RAM, which is unknown today; after the 128 GB upgrade ~110 GB of MoE experts could sit in RAM. |
| Ugreen DXP2800 NAS | No | Storage host. Already runs ollama/open-webui containers; only tiny models (e.g. Whisper on CPU) are realistic. |
| MacBook Neo | Unknown | Specs unknown: machine was offline/unreachable on Nov 14 2026. |
Token/s = theoretical ceiling from memory bandwidth and active weight bytes; not a benchmark. Nothing on this page is a benchmark score.