Gemma 3 12B - NC Bench

google/gemma-3-12b-it

Gemma 3 12B

via OpenRouter

Release Date

Mar 12th, 2025

Context Size

128k

Reasoning

Benchmark Cost

$0.16

Speed

39.2 tok/s

Subcategories

Bad Writing Habits

Detects common prose quality anti-patterns in AI-generated creative writing, including passive voice, past progressive overuse, weak dialogue tags, filter words, purple prose, cliches, AI-ism words/adverbs/names, and more.

Scenario	#1	#2	#3	#4	#5	Total
Detailed Writing Rules
Fantasy: entering an ancient ruin Creative Writing Hallucination	77	76	76	75	70	75%
Horror: alone in an eerie place at night Creative Writing Hallucination	80	80	79	79	71	78%
Literary fiction: old friends reunite Creative Writing Hallucination	81	80	77	76	74	78%
Mystery: examining a crime scene Creative Writing Hallucination	85	82	81	72	71	78%
Romance: separated couple reunites Creative Writing Hallucination	81	79	77	75	74	77%
Thriller: chase through city streets Creative Writing Hallucination	80	78	77	70	70	75%
Detailed Writing Rules						76.79%
genre
Fantasy: entering an ancient ruin Creative Writing Hallucination	79	73	73	72	68	73%
Horror: alone in an eerie place at night Creative Writing Hallucination	81	81	80	80	69	78%
Literary fiction: old friends reunite Creative Writing Hallucination	78	78	74	74	64	74%
Mystery: examining a crime scene Creative Writing Hallucination	81	81	80	76	75	79%
Romance: separated couple reunites Creative Writing Hallucination	78	76	75	74	66	74%
Thriller: chase through city streets Creative Writing Hallucination	80	78	75	75	75	77%
genre						75.66%
Novelcrafter Default Prompt
Fantasy: entering an ancient ruin Creative Writing Hallucination	79	75	74	73	67	74%
Horror: alone in an eerie place at night Creative Writing Hallucination	80	78	76	76	76	77%
Literary fiction: old friends reunite Creative Writing Hallucination	78	75	72	72	72	74%
Mystery: examining a crime scene Creative Writing Hallucination	82	78	74	72	72	76%
Romance: separated couple reunites Creative Writing Hallucination	76	75	75	74	67	74%
Thriller: chase through city streets Creative Writing Hallucination	82	81	79	77	75	79%
Novelcrafter Default Prompt						75.50%
75.98%

Codex Extraction

Evaluates a model's ability to extract structured codex entries (characters, locations, objects, lore) from prose passages and return them as well-formed XML.

Scenario	#1	#2	#3	#4	#5	Total
Long: The Spire of Echoes (Dense) Tooling Reasoning	90	89	89	89	81	88%
Medium: The Hollow (Inferred) Tooling Reasoning	88	87	86	85	3	70%
Medium: Through the Thornveil (Scattered) Tooling Reasoning	85	84	83	83	82	84%
Short: The Rusty Lantern (Explicit) Tooling Reasoning	77	71	70	69	63	70%
77.70%

Codex Red Herring (False Positive Detection)

Tests whether models correctly report "no violations" when a codex is fully consistent with the prose passage. Models that hallucinate false violations (false positives) fail. Uses a 2×2 matrix of text length × codex size, with bare and detailed-entry variants.

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
basic entries
Long text (~1594 words), big codex (51 entries) Hallucination	13	5	4	4	4	4	3	3	3	2	4%
Long text (~1594 words), small codex (11 entries) Hallucination	17	8	8	8	8	8	7	7	7	7	9%
Short text (~524 words), big codex (51 entries) Hallucination	29	7	7	6	6	6	6	5	5	2	8%
Short text (~524 words), small codex (11 entries) Hallucination	7	6	6	5	5	5	5	5	4	3	5%
basic entries											6.47%
detailed entries
Long text (~1594 words), big codex (51 detailed entries) Hallucination	8	8	8	6	6	5	4	4	4	1	5%
Long text (~1594 words), small codex (11 detailed entries) Hallucination	17	13	13	10	10	7	6	6	6	6	9%
Short text (~524 words), big codex (51 detailed entries) Hallucination	29	7	6	6	5	5	5	4	3	3	7%
Short text (~524 words), small codex (11 detailed entries) Hallucination	17	17	13	13	13	13	13	8	7	6	12%
detailed entries											8.43%
7.45%

Codex Violation Detection

Detects factual inconsistencies between a story bible and prose passages. The model must output structured XML identifying each violation with paragraph number and substring.

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
matrix
Large codex (40 entries), long passage (1,019 words) Tooling Reasoning	61	59	55	52	48	44	44	43	40	40	49%
Large codex (40 entries), short passage (165 words) Tooling Reasoning	72	69	69	67	66	66	64	63	63	52	65%
Small codex (7 entries), long passage (734 words) Tooling Reasoning	75	66	66	66	61	61	61	57	57	53	62%
Small codex (7 entries), short passage (165 words) Tooling Reasoning	97	97	97	97	97	97	93	93	93	88	95%
matrix											67.72%
tiers
5 codex entries Tooling Reasoning	93	75	75	75	75	75	75	75	75	75	77%
10 codex entries Tooling Reasoning	92	86	86	86	86	86	79	79	72	64	82%
20 codex entries Tooling Reasoning	65	61	61	61	60	53	52	51	47	3	51%
40 codex entries Tooling Reasoning	55	48	48	48	48	40	40	40	40	33	44%
tiers											63.54%
65.63%

Data extraction

Extract key details from a given block of text.

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
All valid emails Utility	100	100	100	100	100	100	100	100	100	100	100%
Contextual pronoun Utility	100	100	100	100	100	100	100	100	100	100	100%
Fruits excluding citrus Utility	100	100	100	100	100	100	100	100	100	100	100%
Future event time Reasoning	100	100	100	100	100	100	100	100	100	100	100%
Guess the pet Reasoning	100	100	100	100	100	100	100	100	100	100	100%
Highest-rated movie Reasoning	100	100	100	100	100	100	100	100	100	100	100%
Indirect birth year Reasoning	100	100	100	100	100	100	100	100	100	100	100%
What instrument does Lucy play? Reasoning	100	100	100	100	100	100	100	100	100	100	100%
What's the color of the car? Utility	100	100	100	100	100	100	100	100	100	100	100%
What's the correct time? Reasoning	0	0	0	0	0	0	0	0	0	0	0%
Who's the sister? Reasoning	100	100	100	100	100	100	100	100	100	100	100%
Who's the tallest? Reasoning	100	100	100	100	100	100	100	100	100	100	100%
91.67%

Dialogue tags

Various tasks related to dialogue tags in text.

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
dialogue-200
Write 200 words with 10% dialogue Rule Following Utility	97	94	90	83	82	67	44	9	7	3	58%
Write 200 words with 50% dialogue Rule Following Utility	41	4	1	0	0	0	0	0	0	0	5%
Write 200 words with 90% dialogue Rule Following Utility	67	57	51	50	50	50	50	50	44	16	49%
dialogue-200											36.95%
dialogue-500
Write 500 words with 30% dialogue Rule Following Utility	50	49	43	38	30	0	0	0	0	0	21%
Write 500 words with 50% dialogue Rule Following Utility	50	7	0	0	0	0	0	0	0	0	6%
Write 500 words with 70% dialogue Rule Following Utility	61	39	36	34	18	5	3	1	1	0	20%
dialogue-500											15.53%
Ungrouped
Write unattributed dialogue Rule Following	100	100	100	100	100	100	100	100	100	100	100%
36.78%

Language Comprehension

Does the model understand more than just English?

Scenario	#1	#2	#3	#4	#5	Total
Asking for directions (Dutch) Language	0	0	0	0	0	0%
Asking for directions (German) Language	100	100	100	100	100	100%
Friend got new kittens (German) Language	100	100	100	100	0	80%
Friend got new kittens (Tagalog) Language	100	100	100	100	100	100%
70.00%

Language Writing

Can the model generate text in different languages?

Scenario	#1	#2	#3	#4	#5	Total
Character dialogue (French) in a story Language	100	100	100	100	92	98%
Character dialogue (German) in a story Language	100	100	91	91	90	94%
Character dialogue (Hindi) in a story Language	100	81	55	53	38	65%
Character dialogue (Italian) in a story Language	100	100	100	100	89	98%
Character dialogue (Spanish) in a story Language	100	100	100	100	76	95%
90.19%

N-Length Sentences

Write sentences with exactly N words

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
Write sentences with 5 words each Utility	100	100	100	98	96	96	96	96	93	87	96%
Write sentences with 10 words each Utility	95	92	92	90	90	87	87	85	84	74	88%
Write sentences with 20 words each Utility	13	13	9	9	2	2	2	0	0	0	5%
62.93%

Novel outline

Handle questions about the outline of a novel in various formats

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
outline-count
Count acts Utility	100	100	100	100	100	100	100	100	100	100	100%
Count chapters Utility	100	100	100	100	100	100	100	100	100	100	100%
Count scenes Utility	100	0	0	0	0	0	0	0	0	0	10%
outline-count											70.00%
pov-count
Count point of views for Jack and Olivia Utility	100	100	100	100	100	50	50	50	50	50	75%
Count point of views for Jack Harper Utility	100	100	100	100	100	100	100	100	100	100	100%
Count point of views for Olivia Utility	100	0	0	0	0	0	0	0	0	0	10%
pov-count											61.67%
65.83%

Text Replacement

Tests deterministic text transformations: renaming characters/locations, expanding contractions, tense rewriting, POV shifts, gender swaps, combined transformations, and word avoidance. Scored by checking each expected change independently.

Scenario	#1	#2	#3	#4	#5	#6	#7	Total
Generic Prompt
Avoid said/asked/replied/answered Text Editing	100	100	100	100	100	100	100	100%
Character rename: Elena->Mirabel, Gregor->Aldric Text Editing	100	100	100	100	100	100	100	100%
Combined: 3rd person past → 1st person present Text Editing	99	99	98	98	98	98	70	94%
Expand all contractions Text Editing	99	98	98	98	98	98	98	98%
Location rename: market square, outer ring, bridge, northern mines Text Editing	100	100	100	100	100	100	100	100%
Multi-character gender swap: Priya(F)->Rohan(M), Mara unchanged Text Editing	74	74	74	74	74	74	74	74%
Passive voice → active voice Text Editing Hallucination	76	76	76	76	76	76	74	75%
POV shift: 3rd person to 1st person (Elena's perspective) Text Editing	100	100	100	100	100	100	100	100%
Tense rewriting: past to present Text Editing	92	92	92	92	89	81	81	89%
Generic Prompt								92.19%
Specific Prompt
Avoid said/asked/replied/answered Text Editing	100	100	100	100	100	100	89	98%
Character rename: Elena->Mirabel, Gregor->Aldric Text Editing	100	100	100	100	100	100	100	100%
Combined: 3rd person past → 1st person present Text Editing	99	98	98	98	98	98	98	98%
Expand all contractions Text Editing	100	100	100	100	100	100	99	100%
Location rename: market square, outer ring, bridge, northern mines Text Editing	100	100	100	100	100	100	100	100%
Multi-character gender swap: Priya(F)->Rohan(M), Mara unchanged Text Editing	100	89	89	89	89	86	86	90%
Passive voice → active voice Text Editing Hallucination	95	94	94	94	94	93	91	94%
POV shift: 3rd person to 1st person (Elena's perspective) Text Editing	100	100	100	100	100	100	100	100%
Tense rewriting: past to present Text Editing	100	100	100	100	100	100	100	100%
Specific Prompt								97.76%
94.98%

Tool usage within Novelcrafter

Output messages that are related to tool usage within Novelcrafter

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
Create alternate prose sections Tooling	100	100	100	100	100	100	100	100	100	100	100%
100.00%

Voice/dialogue sheets

Extract dialogue from given text as voice sheets.

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
Multiple speakers Rule Following	100	100	100	100	100	0	0	0	0	0	50%
Simple Rule Following	0	0	0	0	0	0	0	0	0	0	0%
Simple (1-shot) 1-shot Rule Following	100	100	100	100	100	100	100	100	100	100	100%
Simple (5-shot) Few-shot Rule Following	100	100	100	100	100	100	100	100	100	100	100%
Unattributed dialogue Rule Following	100	0	0	0	0	0	0	0	0	0	10%
52.00%

Write N of X

Write exactly N words/sentences/paragraphs...

Scenario	#1	#2	#3	#4	#5	#6	#7	#8	#9	#10	Total
paragraphs
1 paragraph summary Utility	100	100	100	100	100	100	100	100	100	100	100%
3 paragraph summary Utility	100	100	100	100	100	100	100	100	100	100	100%
5 paragraph summary Utility	100	100	100	100	100	100	100	100	100	100	100%
paragraphs											100.00%
sentences
1 sentence summary Utility	100	100	100	100	100	100	100	100	100	100	100%
3 sentence summary Utility	100	100	100	100	100	100	100	100	100	100	100%
10 sentence summary Utility	100	100	100	100	100	100	100	100	100	100	100%
20 sentence summary Utility	100	100	100	100	100	100	98	54	2	0	75%
50 sentence summary Utility	98	98	54	0	0	0	0	0	0	0	25%
sentences											80.06%
words
10 word summary Utility	100	100	100	100	100	100	100	100	100	100	100%
20 word summary Utility	100	100	100	100	100	98	98	98	92	77	96%
50 word summary Utility	27	9	2	0	0	0	0	0	0	0	4%
100 word summary Utility	100	92	92	92	92	54	2	2	0	0	53%
200 word summary Utility	100	98	77	9	0	0	0	0	0	0	28%
words											56.25%
75.50%