SEE7/AstheralPlus
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.
AstheralPlus is SEE7's instruction and reasoning model based on GPT-OSS-20B: 20.9 billion total parameters, with approximately 3.6 billion active per token. Mixture-of-experts, 21B total (32 experts, 4 per token). Context: 128K tokens (config). License: apache-2.0.
- Org
- SEE7
- Released
- 2026-11-13
- Status
- just landed · new, unvetted
- License
- apache-2.0 (permissive)
- Parameters
- 20.9B
- Architecture
- MoE · 32 experts, 4/token · GptOssForCausalLM
- Context
- 131,072 tokens (config.json)
- Task
- text-generation · text
- Signal
- 2 HF likes · 214 downloads
Runs on
mixed: real files + estimate; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~17.2 GB: 24GB GPU (3090 / 4090), 24GB GPU (7900 XTX), Mac 24GB
- Q8 ~23.6 GB (estimated size): 32GB GPU (5090), Arc B70 32GB, Mac 36GB, Mac 48GB, CPU only, 32GB RAM (slow)
- FP16 ~43.2 GB (estimated size): 48GB GPU (RTX 6000 Ada / A6000), 96GB GPU (RTX Pro 6000 Blackwell), 1× DGX Spark (128GB), 2× DGX Spark (256GB), Strix Halo 128GB unified (~96GB usable), Mac 64GB (M1 Max), Mac 96GB, Mac 128GB, Mac 192GB, Mac 256GB, Mac 512GB, CPU only, 64GB RAM (slow), CPU only, 128GB RAM (slow), CPU only, 256GB RAM (slow)
Links
First published here 2026-11-13 · updated 2026-11-14 · markdown · JSON