ricursiveAI/Ripple-65B-Preview
main boardmodel
Ripple-65B Base Preview is a pretrain-only checkpoint without any midtraining or synthetic QA data. Dense, 65B params. Context: 2M tokens (config). License: apache-2.0. Vendor-reported: "The base model overall beats Poolside Laguna M.1, and is comparable with DeepSeek V3.1 and V4 Flash on reasoning, math, and code."
- Vendor-reported: The base model overall beats Poolside Laguna M.1, and is comparable with DeepSeek V3.1 and V4 Flash on reasoning, math, and code.
- Org
- ricursiveAI
- Released
- 2026-11-04
- Status
- main board
- License
- apache-2.0 (permissive)
- Parameters
- 64.8B
- Architecture
- dense · RicursiveForCausalLM
- Context
- 2,097,152 tokens (config.json)
- Signal
- 584 HF likes · 10675 downloads · 1 X likes · 1 X post(s)
Runs on
mixed: real files + estimate; need = weights + FP16 KV cache for 8K tokens + 1GB. A sizing calculation, not tested.
- Q4 ~42.6 GB (estimated size): 48GB GPU (RTX 6000 Ada / A6000), Mac 64GB (M1 Max), Mac 96GB, CPU only, 64GB RAM (slow)
- Q8 ~72.5 GB: 96GB GPU (RTX Pro 6000 Blackwell), 1× DGX Spark (128GB), Strix Halo 128GB unified (~96GB usable), Mac 128GB, CPU only, 128GB RAM (slow)
- FP16 ~133.3 GB: 2× DGX Spark (256GB), Mac 192GB, Mac 256GB, Mac 512GB, CPU only, 256GB RAM (slow)
Links
First published here 2026-11-04 · updated 2026-11-06 · markdown · JSON