# EldanRing: winnow

> A Gemma 4 12B fine-tune for typed decisions that also chats and reads images. Ships as GGUF for llama.cpp, holds a 64K context on a 16 GB GPU, and serves /v1/systemone next to /v1/chat/completions.

- Page: https://systemonemodels.tech/eldanring/winnow
- API: https://api.systemonemodels.tech/v1/models/eldanring/winnow
- Download: `pip install systemonemodels && systemone pull eldanring/winnow`

## Facts

| | |
|---|---|
| Maker | EldanRing (https://systemonemodels.tech/eldanring) |
| Decides | choice, score, noul, classify, route |
| Architecture | winnow |
| Base model | google/gemma-4-12b-it |
| Parameters | 12B |
| Context | 64K tokens |
| Licence | apache-2.0 |
| Availability | Open weights |
| Released | 2026-09-20 |
| Latest version | 2026.09 |

## Reported evaluation

Suite: JevBench public subset (231 items), Q8_0. Numbers are the publisher's own.

- Decision accuracy: 85.7%

- Median latency: 143 ms

## Model card

<!-- generated by scripts/seed_catalog.py; edit content/models/catalog.yaml -->

# Winnow-12B

A Gemma 4 12B fine-tune for typed decisions that also chats and reads images. Ships as GGUF for llama.cpp, holds a 64K context on a 16 GB GPU, and serves /v1/systemone next to /v1/chat/completions.

On JevBench's public subset (231 items) the Q8_0 build scores 85.71%, the same as Jev 1.13 measured through OpenRouter on the same inputs, and 81.55% on Kev v9 clean (Jev 87.00%). It is a rank-32 LoRA merged into Gemma 4 12B IT and shipped as BF16 and Q8_0 GGUF with an optional vision projector; on an RTX 5070 Ti the Q8 build holds a 64K context with an image in 15 GiB, and a cached four-question request takes 143 ms. Probabilities use temperature 1.0 with no fitted calibration map. Sibling: Winnow-E4B on Gemma 4 E4B, 80.52% on the same JevBench subset at about twice the decisions per second. Both are Apache-2.0, like Gemma 4.

## What it decides

- **choice** — picks one option from a set
- **score** — places the input on an ordered scale
- **noul** — answers a yes/no question with one calibrated probability
- **classify** — assigns a category from a fixed taxonomy
- **route** — sends the input to one of several destinations

## At a glance

| | |
|---|---|
| Parameters | 12B |
| Base model | `google/gemma-4-12b-it` |
| Maker | EldanRing |
| Released | 2026-09-20 |
| License | apache-2.0 |
| Reported accuracy | 85.7% |
| Reported latency | 143 ms for a cached four-question request on an RTX 5070 Ti (Q8) |

## Get the weights

```bash
pip install systemonemodels
systemone pull eldanring/winnow
```

The files are served from the maker's Hugging Face repository, [`EldanRing/Winnow-12B`](https://huggingface.co/EldanRing/Winnow-12B), and verified against the checksums recorded here.

## Read more

- [Inference server](https://github.com/EldanRing/winnow-inference)
- [Weights on Hugging Face](https://huggingface.co/EldanRing/Winnow-12B)
- [Winnow-E4B](https://huggingface.co/EldanRing/Winnow-E4B)

---

*This page was opened by System One for EldanRing, who can claim the organisation and take it over at any time.*

---

From System One Models — https://systemonemodels.tech/ · every System One model: https://systemonemodels.tech/system-one-models