Gemma 3 27B - NC Bench

google/gemma-3-27b-it

Gemma 3 27B

via OpenRouter

Release Date

Mar 12th, 2025

Context Size

128k

Reasoning

Benchmark Cost

$0.25

Speed

30.0 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	80	78	77	76	75	77%
Horror: alone in an eerie place at night Creative Writing Hallucination	84	83	79	76	76	80%
Literary fiction: old friends reunite Creative Writing Hallucination	82	82	80	75	74	79%
Mystery: examining a crime scene Creative Writing Hallucination	91	86	82	81	79	84%
Romance: separated couple reunites Creative Writing Hallucination	82	81	78	73	73	77%
Thriller: chase through city streets Creative Writing Hallucination	86	85	83	79	74	81%
Detailed Writing Rules						79.70%
genre
Fantasy: entering an ancient ruin Creative Writing Hallucination	79	71	71	71	67	72%
Horror: alone in an eerie place at night Creative Writing Hallucination	85	78	75	75	73	77%
Literary fiction: old friends reunite Creative Writing Hallucination	80	79	76	74	72	76%
Mystery: examining a crime scene Creative Writing Hallucination	85	83	80	77	73	79%
Romance: separated couple reunites Creative Writing Hallucination	82	79	78	78	76	79%
Thriller: chase through city streets Creative Writing Hallucination	88	85	77	74	74	80%
genre						77.14%
Novelcrafter Default Prompt
Fantasy: entering an ancient ruin Creative Writing Hallucination	76	75	74	73	72	74%
Horror: alone in an eerie place at night Creative Writing Hallucination	84	82	80	80	79	81%
Literary fiction: old friends reunite Creative Writing Hallucination	84	83	81	79	76	81%
Mystery: examining a crime scene Creative Writing Hallucination	84	82	80	80	77	81%
Romance: separated couple reunites Creative Writing Hallucination	84	82	80	80	78	81%
Thriller: chase through city streets Creative Writing Hallucination	86	83	81	81	79	82%
Novelcrafter Default Prompt						79.86%
78.90%

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	91	90	88	88	85	89%
Medium: The Hollow (Inferred) Tooling Reasoning	93	92	91	88	88	91%
Medium: Through the Thornveil (Scattered) Tooling Reasoning	95	94	93	91	90	93%
Short: The Rusty Lantern (Explicit) Tooling Reasoning	92	90	88	88	84	88%
90.01%

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	6	6	6	5	5	5	4	4	3	2	4%
Long text (~1594 words), small codex (11 entries) Hallucination	10	10	10	10	8	8	8	8	7	5	9%
Short text (~524 words), big codex (51 entries) Hallucination	6	6	6	6	5	5	5	5	5	5	5%
Short text (~524 words), small codex (11 entries) Hallucination	13	10	10	8	8	8	7	6	6	6	8%
basic entries											6.61%
detailed entries
Long text (~1594 words), big codex (51 detailed entries) Hallucination	7	4	4	3	3	3	3	3	2	2	3%
Long text (~1594 words), small codex (11 detailed entries) Hallucination	10	8	8	8	7	7	7	7	6	5	7%
Short text (~524 words), big codex (51 detailed entries) Hallucination	6	6	5	5	5	5	5	5	5	4	5%
Short text (~524 words), small codex (11 detailed entries) Hallucination	13	10	8	8	7	7	6	6	6	5	8%
detailed entries											5.86%
6.23%

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	71	68	63	57	56	55	53	53	52	44	57%
Large codex (40 entries), short passage (165 words) Tooling Reasoning	78	77	76	75	72	72	72	72	72	71	74%
Small codex (7 entries), long passage (734 words) Tooling Reasoning	84	83	83	80	80	75	75	71	71	71	77%
Small codex (7 entries), short passage (165 words) Tooling Reasoning	90	88	88	84	84	84	84	80	80	80	84%
matrix											73.20%
tiers
5 codex entries Tooling Reasoning	83	83	83	83	83	83	83	83	83	83	83%
10 codex entries Tooling Reasoning	100	100	92	86	86	86	86	86	86	72	88%
20 codex entries Tooling Reasoning	82	77	77	73	73	69	69	69	69	69	73%
40 codex entries Tooling Reasoning	87	87	84	80	80	80	80	79	75	75	81%
tiers											81.31%
77.26%

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	43	41	7	3	0	0	0	0	0	0	9%
Write 200 words with 50% dialogue Rule Following Utility	10	0	0	0	0	0	0	0	0	0	1%
Write 200 words with 90% dialogue Rule Following Utility	50	50	50	49	49	49	45	39	18	18	42%
dialogue-200											17.38%
dialogue-500
Write 500 words with 30% dialogue Rule Following Utility	49	3	0	0	0	0	0	0	0	0	5%
Write 500 words with 50% dialogue Rule Following Utility	0	0	0	0	0	0	0	0	0	0	0%
Write 500 words with 70% dialogue Rule Following Utility	1	0	0	0	0	0	0	0	0	0	0%
dialogue-500											1.79%
Ungrouped
Write unattributed dialogue Rule Following	100	100	100	100	100	100	100	100	100	100	100%
22.50%

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	0	0	60%
Friend got new kittens (German) Language	100	100	100	100	100	100%
Friend got new kittens (Tagalog) Language	100	100	100	100	100	100%
65.00%

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	93	99%
Character dialogue (German) in a story Language	100	100	100	92	83	95%
Character dialogue (Hindi) in a story Language	100	87	50	50	50	67%
Character dialogue (Italian) in a story Language	93	93	92	86	85	90%
Character dialogue (Spanish) in a story Language	100	100	94	94	93	96%
89.42%

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	100	100	98	98	96	96	93	98%
Write sentences with 10 words each Utility	95	92	92	92	92	90	90	87	84	83	90%
Write sentences with 20 words each Utility	60	56	53	47	44	28	17	2	2	2	31%
72.87%

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	0	0	0	0	0	0	0	0	0	0	0%
outline-count											66.67%
pov-count
Count point of views for Jack and Olivia Utility	100	100	50	50	50	50	0	0	0	0	40%
Count point of views for Jack Harper Utility	0	0	0	0	0	0	0	0	0	0	0%
Count point of views for Olivia Utility	100	100	100	0	0	0	0	0	0	0	30%
pov-count											23.33%
45.00%

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	99	98	98	98	94	98%
Expand all contractions Text Editing	100	100	100	100	99	99	98	100%
Location rename: market square, outer ring, bridge, northern mines Text Editing	100	100	100	100	96	96	96	98%
Multi-character gender swap: Priya(F)->Rohan(M), Mara unchanged Text Editing	98	98	93	93	93	93	93	94%
Passive voice → active voice Text Editing Hallucination	77	77	77	77	77	75	75	76%
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	96	91	91	85	84	81	80	87%
Generic Prompt								94.85%
Specific Prompt
Avoid said/asked/replied/answered Text Editing	100	100	100	100	100	100	97	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	99	99	99	99	97	99%
Expand all contractions Text Editing	100	100	100	100	100	100	100	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	100	100	100	100	100	100	100%
Passive voice → active voice Text Editing Hallucination	94	93	92	91	79	77	77	86%
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	99	100%
Specific Prompt								98.25%
96.55%

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	0	0	0	0	0	0	0	0	0	0	0%
Simple Rule Following	100	100	0	0	0	0	0	0	0	0	20%
Simple (1-shot) 1-shot Rule Following	0	0	0	0	0	0	0	0	0	0	0%
Simple (5-shot) Few-shot Rule Following	100	100	100	0	0	0	0	0	0	0	30%
Unattributed dialogue Rule Following	100	100	100	100	100	100	100	100	100	100	100%
30.00%

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	98	98	98	92	54	9	85%
50 sentence summary Utility	100	100	100	100	100	100	100	92	77	54	92%
sentences											95.44%
words
10 word summary Utility	100	100	100	100	100	100	100	100	100	100	100%
20 word summary Utility	100	98	98	92	92	92	77	77	27	27	78%
50 word summary Utility	100	100	100	98	98	98	92	77	77	9	85%
100 word summary Utility	98	92	27	27	9	0	0	0	0	0	25%
200 word summary Utility	27	27	2	0	0	0	0	0	0	0	6%
words											58.88%
82.43%