November 2026 · Independent humanizer benchmarks. Explore the evidence
← All humanizers

Humanizer 12B · OptiQ 4-bit

Verified by HumanizeBenchWe ran this model ourselves on the benchmark source examples and checked its execution records. Verification confirms the tested model’s provenance; it does not guarantee performance or imply that every detector returned a result.
Open weights33 saved rewrites

Humanizer 12B · OptiQ 4-bit has a current composite of 43.13/100 in our shared November evaluation. Readability is 77.1 and rescaled meaning preservation is 60.9. 231 of 231 detector measurements are available.

This entry is the requested mlx-optiq-4bit artifact from mlx-community/humanizer-12b-OptiQ-4bit. Related formats are separate roster entries, not necessarily independent underlying models.

Overall score43.13/100
Panel bypass score36.3%
Readability77.1%
Prompts tested33

Detector human-likelihood scores

Mean reported human score on measured outputs. Missing results stay missing. Coverage is shown beside each detector.

GPTZero
14.933/33
ZeroGPT
16.033/33
Copyleaks
48.533/33
Pangram
35.733/33
Grammarly
73.433/33
QuillBot
99.733/33

Performance by category

HumanizeBench v2 per-category values.

Academic essay51.1Application essay37.4Blog post50.5Business email36.3Marketing copy62.4Discussion board66.3News article46.0

Quality and penalties

0.9022Raw meaning cosine
0/33Identical outputs
0/33Refusals
7.0Penalty points
RuleOccurrencesDeduction
Meaning drift7−7

Artifact & evaluation record

Entry
mlx-community/humanizer-12b-OptiQ-4bit
Source
Recorded GPU generation
Family
jialinyyzz-humanizer
Pinned revision
9b992be61bdd7516a022c2597a47541cdbf6a340
Prompt basis
Pinned model card inference example; seed 7 and unspecified sampling filters frozen before any inference; no detector tuning
Scoring
56.5% bypass · 24% meaning · 12% readability · 7.5% consistency, minus fixed penalties.
Measurement notes
NVIDIA quality substitutes; current Polygraph routes; no historical-score fallback.

Inspect the recorded text

The source and rewritten text are available for every sample.

SampleCategorySource modelInput wordsReadabilityDetector scores
S01 Academic essayAcademic essaygpt-5.6-terra31573.37/7
S02 Academic essayAcademic essayclaude-sonnet-529680.07/7
S03 Application essayApplication essaygemini-3.5-flash29180.07/7
S04 Application essayApplication essaygpt-5.6-luna27180.07/7
S05 Blog postBlog postclaude-haiku-4-5-2025100128080.07/7
S06 Blog postBlog postgemini-2.5-flash32980.07/7
S07 Business emailBusiness emailgpt-5.6-terra26280.07/7
S08 Marketing copyMarketing copyclaude-sonnet-529980.07/7
S09 Marketing copyMarketing copygemini-3.5-flash19178.07/7
S10 Discussion boardDiscussion boardgpt-5.6-luna26880.07/7
S11 News articleNews articleclaude-haiku-4-5-2025100130078.07/7
S12 Academic essayAcademic essaygemini-2.5-flash25975.37/7
S13 Academic essayAcademic essaygpt-5.6-terra34978.07/7
S14 Application essayApplication essayclaude-sonnet-531483.37/7
S15 Application essayApplication essaygemini-3.5-flash29275.37/7
S16 Blog postBlog postgpt-5.6-luna31380.07/7
S17 Blog postBlog postclaude-haiku-4-5-2025100128172.07/7
S18 Business emailBusiness emailgemini-2.5-flash18880.07/7
S19 Marketing copyMarketing copygpt-5.6-terra25772.77/7
S20 Marketing copyMarketing copyclaude-sonnet-519172.77/7
S21 Discussion boardDiscussion boardgemini-3.5-flash25163.37/7
S22 News articleNews articlegpt-5.6-luna29872.07/7
S23 Academic essayAcademic essayclaude-haiku-4-5-2025100131268.77/7
S24 Academic essayAcademic essaygemini-2.5-flash25876.77/7
S25 Application essayApplication essaygpt-5.6-terra30080.07/7
S26 Application essayApplication essayclaude-sonnet-532980.77/7
S27 Blog postBlog postgemini-3.5-flash29880.07/7
S28 Blog postBlog postgpt-5.6-luna32376.07/7
S29 Business emailBusiness emailclaude-haiku-4-5-2025100122880.07/7
S30 Marketing copyMarketing copygemini-2.5-flash27778.07/7
S31 Marketing copyMarketing copygpt-5.6-terra19568.77/7
S32 Discussion boardDiscussion boardclaude-sonnet-525282.07/7
S33 News articleNews articlegemini-3.5-flash29581.37/7