BeriaLabs/Qwen-3.8-gollum
just landed · new, unvettedmodel
New, unvetted: created or first seen in the last few days, not from a trusted org and not past the signal threshold yet.
27B params. License: apache-2.0. Vendor-reported: "Contra el rival directo: paridad estadística en los dos ejes de rechazo (las diferencias caen dentro de los IC95), con gollum nominalmente primero en unsafe y 3,7× más conciso."
- Vendor-reported: Contra el rival directo: paridad estadística en los dos ejes de rechazo (las diferencias caen dentro de los IC95), con gollum nominalmente primero en unsafe y 3,7× más conciso.
- Org
- BeriaLabs
- Released
- 2026-10-13
- Status
- just landed · new, unvetted
- License
- apache-2.0 (permissive)
- Parameters
- 26.9B
- Signal
- 35 HF likes · 38 downloads
- Weights
- gated
Runs on
mixed: real files + estimate; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~19.4 GB (estimated size): 24GB GPU (3090 / 4090), 32GB GPU (5090), 24GB GPU (7900 XTX), Mac 36GB, CPU only, 32GB RAM (slow)
- Q8 ~31.9 GB (estimated size): 48GB GPU (RTX 6000 Ada / A6000), Mac 48GB, Mac 64GB (M1 Max), CPU only, 64GB RAM (slow)
- FP16 ~57.1 GB: 96GB GPU (RTX Pro 6000 Blackwell), 1× DGX Spark (128GB), 2× DGX Spark (256GB), Strix Halo 128GB unified (~96GB usable), Mac 96GB, Mac 128GB, Mac 192GB, Mac 256GB, Mac 512GB, CPU only, 128GB RAM (slow), CPU only, 256GB RAM (slow)
Links
First published here 2026-10-13 · updated 2026-10-31 · markdown · JSON