---
title: MachinePulseAI/karpo-decision-31B
url: https://recentmodel.com/models/MachinePulseAI/karpo-decision-31B/
status: just landed · new, unvetted
published: 2026-11-05T15:33:15Z
updated: 2026-11-14T00:35:18Z
source: https://huggingface.co/MachinePulseAI/karpo-decision-31B
---

# MachinePulseAI/karpo-decision-31B

karpo-decision-31B is a modified version of google/gemma-4-31B-it (Apache-2.0): decoder weights fine-tuned, readout added. Dense, 31B params. Context: 256K tokens (config). License: machinepulse-community.


- **Org:** MachinePulseAI
- **Released:** 2026-11-05
- **Status:** just landed · new, unvetted
- **License:** machinepulse-community (restricted)
- **Parameters:** 31.3B
- **Architecture:** dense · Gemma4ForConditionalGeneration
- **Context:** 262,144 tokens (config.json)
- **Task:** text-classification · text
- **Signal:** 7 HF likes · 496 downloads

## Runs on

mixed: real files + estimate; sizing calculation, not tested.

- Q4 ~27.9 GB (estimated): 32GB GPU (5090), Mac 48GB
- Q8 ~42.3 GB (estimated): 48GB GPU (RTX 6000 Ada / A6000), Mac 64GB (M1 Max), CPU only, 64GB RAM (slow)
- FP16 ~71.7 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

- [Hugging Face](https://huggingface.co/MachinePulseAI/karpo-decision-31B)