YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/model-cards#model-card-metadata)

Glass Benchmark

Logic puzzles a model must answer correctly. In order to even take the Glass Tests, a model must be sufficiently uncensored with few to no refusals, and support at least 16K context.

Glass A Test

Pass

BeaverAI/fallen-gpt-oss-20B-v1b-GGUF
coder3101/Cydonia-24B-v4.3-heretic
TheDrummer/Fallen-Command-A-111B-v1

Fail

BeaverAI/Fallen-Mistral-Small-3.1-24B-v1e
BeaverAI/Fallen-QwQ-32B-v1d-GGUF
dphn/Dolphin-Mistral-24B-Venice-Edition
huihui-ai/Huihui-GLM-4.5-Air-abliterated-GGUF
mlabonne/gemma-3-27b-it-abliterated
Naphula/MN-12B-Mag-Mell-R1-Uncensored
TheDrummer/Precog-24B-v1

Glass B Test

Pass

Naphula/MN-12B-Mag-Mell-R1-Uncensored

Fail

BeaverAI/fallen-gpt-oss-20B-v1b-GGUF
BeaverAI/Fallen-Mistral-Small-3.1-24B-v1e
BeaverAI/Fallen-QwQ-32B-v1d-GGUF
coder3101/Cydonia-24B-v4.3-heretic
dphn/Dolphin-Mistral-24B-Venice-Edition
huihui-ai/Huihui-GLM-4.5-Air-abliterated-GGUF
mlabonne/gemma-3-27b-it-abliterated
TheDrummer/Fallen-Command-A-111B-v1
TheDrummer/Precog-24B-v1

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support