# GameDecider by Holdout Labs

> GameDecider scores English game scenes with an Ettin-400M model. Check private-test limits, abstention coverage, ONNX files and runtime guards.

Updated 2026-11-12 (Australia/Adelaide)

## Quick answers

### What is GameDecider by Holdout Labs?

[GameDecider](<https://decisionmodels.ai/models/gamedecider>): An English game-scene decision model derived from Ettin-400M, with ONNX files and a Python runtime.

[GameDecider facts](<https://decisionmodels.ai/models/gamedecider>)

[Source](<https://huggingface.co/holdoutlabs/gamedecider>)

### Who makes GameDecider?

[GameDecider](<https://decisionmodels.ai/models/gamedecider>): Holdout Labs

[GameDecider facts](<https://decisionmodels.ai/models/gamedecider>)

[Source](<https://huggingface.co/holdoutlabs/gamedecider>)

By Holdout Labs · Hugging Face weight release: 2026-11-09 · OPEN

An English game-scene decision model derived from Ettin-400M, with ONNX files and a Python runtime. It scores NPC intent, mood and quest questions, with optional abstention and 42 compiled adapters. The runtime adds option-order averaging and a minor-safety guard; raw ONNX does not include either.

## Facts

- Inputs: Text or JSON game scenes; choice with 2 to 32 options, score with 2 to 10 levels, and yes/no
- Limits: 3,072 state tokens; two-option choices and choices with 11 or more options never abstain under the published profiles
- Package: 792.5 MB default fp16-weight ONNX file; Python runtime 0.1.0
- Licence: Apache-2.0; upstream Ettin-400M is MIT
- Scope: Fictional game situations; the card excludes consequential decisions about real people

## Claims and sources

- **VENDOR-REPORTED**: The maker reports 60.0% accuracy on 2,993 private questions from unseen task families with option-order averaging. Default thresholds cover 66.2% of that set, versus 97.4% of dev questions. The private evaluation sets and training rows are not released. [source](<https://huggingface.co/holdoutlabs/gamedecider>)
- **VENDOR-REPORTED**: On an AMD Ryzen 5 3600X, the maker reports warm model p50 0.92 s for one question in a single option order and 1.80 s with default choice-order averaging. The 200 held-out rows have a median 325 tokens. [source](<https://huggingface.co/holdoutlabs/gamedecider>)

## Links

- [Source: Card and files](<https://huggingface.co/holdoutlabs/gamedecider>)
- [Source: Training run record](<https://huggingface.co/holdoutlabs/gamedecider/blob/main/model/run.json>)
- [Source: Data sources](<https://huggingface.co/holdoutlabs/gamedecider/blob/main/DATA_SOURCES.md>)
- [Source: Licence](<https://huggingface.co/holdoutlabs/gamedecider/blob/main/LICENSE>)

## Related projects

No related projects are listed in the registry.