NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37PES4F9ADV7ZECCXS9EWB0
Run Details
Model:
Claude Opus 5.5 (Reasoning)
Test:
Bad Writing Habits
Scenario:
Mystery: examining a crime scene
Duration:
53.0s
Cost:
$0.081668
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
100.00%
Adverbs in dialogue tags
(details)
91.16%
AI-ism adverb frequency
(details)
100.00%
AI-ism character names
(details)
100.00%
AI-ism location names
(details)
79.36%
AI-ism word frequency
(details)
100.00%
Cliché density
(details)
100.00%
Emotion telling (show vs. tell)
(details)
91.38%
Filter word density
(details)
100.00%
Gibberish response detection
(details)
100.00%
Markdown formatting overuse
(details)
100.00%
Missing dialogue indicators (quotation marks)
(details)
100.00%
Name drop frequency
(details)
100.00%
Narrator intent-glossing
(details)
100.00%
"Not X but Y" pattern overuse
(details)
100.00%
Overuse of "that" (subordinate clause padding)
(details)
100.00%
Paragraph length variance
(details)
98.94%
Passive voice overuse
(details)
100.00%
Past progressive (was/were + -ing) overuse
(details)
100.00%
Em-dash & semicolon overuse
(details)
100.00%
Purple prose (modifier overload)
(details)
100.00%
Repeated phrase echo
(details)
100.00%
Sentence length variance
(details)
72.62%
Sentence opener variety
(details)
70.92%
Adverb-first sentence starts
(details)
58.30%
Pronoun-first sentence starts
(details)
61.06%
Subject-first sentence starts
(details)
0.00%
Subordinate conjunction sentence starts
(details)
100.00%
Technical jargon density
(details)
100.00%
Useless dialogue additions
(details)
100.00%
Dialogue tag variety (said vs. fancy)
(details)
90.7913%
Overall
Model input
System
You are an expert novelist with a gift for vivid, immersive prose. You write in tight third-person limited point of view with a strong narrative voice. Your prose is clean, precise, and engaging — you favor strong verbs, concrete details, and natural-sounding dialogue. You will be given a story bible (codex) containing canonical details about characters, locations, items, and lore. Use these details to ground your writing in the established world. Stay consistent with the codex. Write only the prose of the scene. Do not include titles, chapter headings, markdown formatting, or any commentary before or after the scene.
User/Human
<codex> <character id="4416" name="Detective Harlow Quinn" aliases="Quinn, Detective Quinn, Harlow" pronouns="she/her" species="Human" storyRole="Antagonist"> <eyeColor>brown</eyeColor> <hair>closely cropped salt-and-pepper</hair> <height>5'9"</height> <age>41</age> <jawline>sharp jaw</jawline> <watch>worn leather watch on her left wrist</watch> <bearing>military precision</bearing> <description>A Metropolitan Police detective with 18 years of decorated service who suspects the clique is involved in criminal activity.</description> <backstory>She lost her partner, DS Morris, during a case three years ago under unexplained circumstances that had supernatural origins she doesn't yet understand.</backstory> </character> <character id="4414" name="Eva Kowalski" aliases="Eva, Evie" pronouns="she/her" species="Human" storyRole="Best Friend"> <eyeColor>green</eyeColor> <hair>curly red</hair> <height>5'4"</height> <age>26</age> <glasses>round glasses</glasses> <satchel>worn leather satchel full of books</satchel> <freckles>freckled complexion</freckles> <nervousHabit>tucking hair behind her left ear</nervousHabit> <description>Aurora's childhood friend and an occult researcher.</description> <backstory>Moved to London two years before Aurora and works as a research assistant at the British Museum's restricted archives. She holds a master's degree in Ancient History from Oxford.</backstory> </character> <location id="856" name="The Veil Market" aliases="Veil Market, the Market, underground market" realm="Earth"> <currentLocation>abandoned Tube station beneath Camden</currentLocation> <entryRequirement>bone token</entryRequirement> <schedule>moves locations every full moon</schedule> <description>A hidden supernatural black market that sells enchanted goods, banned alchemical substances, and information.</description> </location> <item id="2004" name="Veil Compass" aliases="the Compass, shadow compass" type="Tool" origin="The Veil Market"> <casingMaterial>small brass compass</casingMaterial> <faceMarkings>face etched with protective sigils</faceMarkings> <needle>points toward the nearest supernatural rift or portal</needle> <patina>casing has a patina of verdigris</patina> <creator>crafted by a Shade artisan</creator> <description>A compass attuned to supernatural energy.</description> </item> </codex> Write a fully fleshed out scene of at least 1000 words in the mystery genre. A detective arrives at a crime scene in an unusual location. Something is wrong — the evidence doesn't add up. A colleague is present, offering a different interpretation. Through careful observation and deduction, the detective begins to see what others have missed. Use the codex entries above as your world reference for characters and setting. Only include the prose of the scene without any markdown or commentary.
Model response
Messages
Meta
AI
The stairs went down farther
than they should have.
Harlow Quinn counted them without
meaning to, a habit from her Army years that had never quite let go. Ninety-four steps, iron-edged and slick with a century of damp, spiralling past tiles the colour of old teeth.
Somewhere around sixty the air
changed. It stopped smelling of Camden, of diesel and fried onions and the canal, and started smelling of nothing at all.
That bothered her more than
rot would have. At the bottom a uniformed constable lifted the tape for her.
He looked about nineteen and
grey around the mouth. "Ma'am. DS Reyes is on the platform." "How long have you been down here?" "Since four, ma'am." He hesitated. "It's cold, isn't it? For August." It was.
Her breath hung in front
of her when she answered him, and she
could feel
the
chill
working through the wool of her coat.
She checked her watch, the
worn leather strap stiff against her left wrist. 6:52 a.m.
She filed the temperature away
and walked on.
The platform opened in front
of her like the inside of a lung.
Arc lamps on tripods threw
hard white light across the curved ceiling, and the tunnel mouths at either end swallowed it whole.
The station had been sealed
since the war, according to the TfL file on her phone.
Nobody had come down here
in eighty years except rats and the occasional urban explorer with a GoPro and more nerve than sense.
The dead man lay near
the platform edge, on his back, arms at his sides.
Callum Reyes stood over him
with a coffee in one gloved hand and an expression of settled boredom.
He was younger than Quinn
by a decade, good at his job in the way that people are good at jobs they have decided not to find interesting. "Morning, guv. Sorry to drag you out." "You didn't. The Super did."
She stopped two metres from
the body and did not go closer. "Talk." "White male, forties. No ID, no phone, no wallet. Explorer found him about three-thirty and nearly had a coronary. Doc's had a first
look
. No obvious trauma, no ligature marks, nothing under the nails." Reyes shrugged. "My money's on an overdose. He's come down to use, or somebody's brought him down to party, it's gone wrong, and his mates have legged it and taken his stuff. We get a couple of these a year in the old stations." "Down ninety-four steps." "People do strange things when they're high." Quinn crouched.
The victim wore a charcoal
suit, well cut, and the kind of brogues that cost more than a constable made in a week. His face was calm.
His eyes were open, and
there was a thin rime of frost in the lashes. "Doc say anything about that?"
Reyes followed her look and
his mouth tightened, just
slightly
. "Said it was odd. Said the core temperature was odd, too. Lower than it should be for the time frame."
He waved the coffee at
the tunnel. "It's cold down here. Draughts." "It's cold," Quinn agreed. "It isn't freezing."
She stood and turned a
slow circle, letting her eyes adjust to what the lamps showed and what they didn't. The first thing she noticed was the dust.
It lay thick across the
platform, grey and soft as ash, and in the arc light it took footprints the way snow did.
She could see her own
trail coming in from the stairs, and the constable's, and Reyes's, and the scuffed arc where the doctor had knelt. Four sets. Five, counting the explorer, whose trainers had left a frantic zigzag that ended at the body and fled back again. There were no other prints near the dead man. None. "How did he get here, Callum?" "Walked." "Show me." Reyes looked down. She watched him
see
it, watched him
start to
build the explanation before he'd fully understood the problem. "He's been carried. Or he came in from the tunnel side." "The tunnel side is a two-foot drop onto the track bed. There's dust on his trousers, but only on the back, where he's lying in it. There's none on the knees or the cuffs."
She pointed with her chin
at the brogues. "And
look
at his soles." The soles were clean.
Not polished, but clean, the way shoes were
when a man had walked on pavement and carpet and nothing else. No grey on them at all. "So he was carried," Reyes said. "Then where are the carrier's footprints?"
He didn't have an answer
for that. He drank some coffee instead. Quinn moved away from the body and walked the length of the platform, keeping close to the wall. Out here, beyond the ring of disturbed dust around the crime scene, the surface was smooth, almost
pristine
. Almost. Every few metres she found a small round mark, a dimple in the grey the size of a two-pence piece, with a drip of something hard and pale at its centre.
She crouched and touched one
with a gloved fingertip. Wax. Tallow, by the smell of it, animal fat, the kind nobody had burned in London for a hundred years. The dimples ran in a line.
Then another line, parallel, three
metres further out. She paced it. The rows ran the whole length of the platform, evenly spaced, like the outlines of stalls at a street market. Like somebody had set up shop down here, with hundreds of candles, and then lifted every trace of themselves out of the dust except the candle drips. A memory rose, uninvited: Morris on a stairwell in Deptford three years ago, calling up to her that the door at the bottom was cold to the touch. The last thing he'd said to her. The inquest had called it misadventure. Quinn had never once believed it.
She put the memory back
where she kept it and kept walking. "Guv." Reyes had followed her. "The doc found something in his hand. I bagged it. Didn't think it was relevant." He held out an evidence bag. Inside was a compass, small and brass, the casing gone green at the seams with verdigris. Someone had
etched
the face with symbols, dense looping marks that weren't letters in any alphabet Quinn recognised. They looked less like decoration than like instructions. "Antique," Reyes said. "Probably nicked. Probably what he was going to sell for his next fix." Quinn turned the bag in the light. The needle didn't point north. She knew roughly where north was; the stairs had come down facing it. The needle pointed across the platform, toward the far wall, and when she stepped left, it stayed fixed on the same spot. When she stepped right, it swung to keep it. "Callum. Which way's north?" He pointed. Correctly. She held up the compass so he could
see
the needle pointing the other way. He looked at it for a long moment and said, "It's broken." "It's not broken. It's pointing at something." She followed it. The far wall
was tiled
like the rest, cream and oxblood, with the station's name picked out in a
mosaic
panel that had lost half its letters. The needle led her to a patch just to the left of the panel, a stretch of tile no different from the tiles around it. Except that when she held her palm an inch away, she felt heat coming off it. Not
warmth
. Heat, like standing near a radiator, in a station where her breath still smoked. She took her glove off and touched the tile. It was smooth and almost hot, and beneath her fingers,
very
faintly, she felt a vibration, the way a window hums when a lorry goes past outside. There was no lorry. There was nothing above them but eighty feet of London clay. "Guv?" "Get the explorer back in. I want to know exactly what time he came down, and I want to know if he saw lights." "Lights?" "Candles. People. Anything." She pulled her glove back on, flexing her fingers against the lingering heat. "And find out when the last full moon was." Reyes gave her a
look
she'd seen before, from colleagues, from her Superintendent, from the Federation rep after Deptford. "Harlow, with respect—" "This man didn't walk in here and he wasn't carried in. Somebody ran a business on this platform last night, a big one, and cleaned up behind them better than any forensics team I've ever worked with. He died in the middle of it, and they left him because they couldn't take him where they were going." She looked back at the wall. "And they went through there." "That's a wall." "Then why is it warm?" He didn't answer. She walked back to the body. Something else was nagging at her, and she crouched again and looked at the dead man's suit, at the inside breast pocket, which sat
slightly
open. Reyes had said nothing on him. Reyes was usually thorough, but Reyes had decided this was an overdose before he'd even seen the brogues. With two fingers she drew the pocket wider. Tucked into the lining, where the stitching had come loose, was a slip of card. She eased it out. It was a reading-room request slip, the kind the British Museum used for its restricted collections. The requested item was a catalogue number she didn't recognise. Under *
Authorised by
* someone had signed in quick, looping ballpoint, the letters slanting hard to the right. *E. Kowalski.* Quinn
knew that name
. She had written it in her notebook six weeks ago, in a list of five names, beside a photograph of a curly-haired redhead in round glasses tucking her hair behind her ear on the steps of a Camden pub while her friends laughed around her. Behind her, on the far wall, she thought the tiles hummed a little louder. She slid the slip into an evidence bag and sealed it, and for the first time since coming down those ninety-four steps, Harlow Quinn felt something warmer than the cold. It wasn't comfort. It was the old, clean feeling of a thread pulled taut.
created_at
"2026-09-23T17:54:14.816Z"
9/23/2026, 5:54:14 PM
model
"anthropic/claude-opus-5.5-20260921"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4771
moderation_latency
(null)
generation_time
52948
tokens_prompt
896
tokens_completion
2642
native_tokens_prompt
1287
native_tokens_completion
3826
native_tokens_completion_images
(null)
native_tokens_reasoning
661
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"end_turn"
service_tier
"default"
usage
0.081668
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790186054-Wk9R2V8JA4iktTtfDX3p"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790186054-01VmI5Z43MXBkrgA8u8Q"
upstream_id
"msg_011CfLoL4RxNB4HcSSZPKkYy"
provider_responses
0
endpoint_id
"3a2388bc-3740-4e64-a1f5-4e301726b6b9"
id
"msg_011CfLoL4RxNB4HcSSZPKkYy"
is_byok
false
latency
1241
model_permaslug
"anthropic/claude-opus-5.5-20260921"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.081668
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Claude Platform on AWS"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Adverbs in dialogue tags
Target: ≤10% dialogue tags with adverbs
totalTags
12
adverbTagCount
0
adverbTags
(empty)
dialogueSentences
41
tagDensity
0.293
leniency
0.585
rawRatio
0
effectiveRatio
0
91.16%
AI-ism adverb frequency
Target: <2% AI-ism adverbs (58 tracked)
wordCount
1696
totalAiIsmAdverbs
3
found
0
adverb
"slightly"
count
2
1
adverb
"very"
count
1
highlights
0
"slightly"
1
"very"
100.00%
AI-ism character names
Target: 0 AI-default names (17 tracked, −20% each)
codexExemptions
(empty)
found
(empty)
100.00%
AI-ism location names
Target: 0 AI-default location names (33 tracked, −20% each)
codexExemptions
(empty)
found
(empty)
79.36%
AI-ism word frequency
Target: <2% AI-ism words (290 tracked)
wordCount
1696
totalAiIsms
7
found
0
word
"could feel"
count
1
1
word
"chill"
count
1
2
word
"pristine"
count
1
3
word
"the last thing"
count
1
4
word
"etched"
count
1
5
word
"mosaic"
count
1
6
word
"warmth"
count
1
highlights
0
"could feel"
1
"chill"
2
"pristine"
3
"the last thing"
4
"etched"
5
"mosaic"
6
"warmth"
100.00%
Cliché density
Target: ≤1 cliche(s) per 800-word window
totalCliches
0
maxInWindow
0
found
(empty)
highlights
(empty)
100.00%
Emotion telling (show vs. tell)
Target: ≤3% sentences with emotion telling
emotionTells
0
narrationSentences
111
matches
(empty)
91.38%
Filter word density
Target: ≤3% sentences with filter/hedge words
filterCount
3
hedgeCount
1
narrationSentences
111
filterMatches
0
"watch"
1
"look"
2
"see"
hedgeMatches
0
"start to"
100.00%
Gibberish response detection
Target: ≤1% gibberish-like sentences (hard fail if a sentence exceeds 800 words)
analyzedSentences
140
gibberishSentences
0
adjustedGibberishSentences
0
longSentenceCount
0
runOnParagraphCount
0
giantParagraphCount
0
wordSaladCount
0
repetitionLoopCount
0
controlTokenCount
0
repeatedSegmentCount
0
maxSentenceWordsSeen
63
ratio
0
matches
(empty)
100.00%
Markdown formatting overuse
Target: ≤5% words in markdown formatting
markdownSpans
2
markdownWords
4
totalWords
1696
ratio
0.002
matches
0
"Authorised by"
1
"E. Kowalski."
100.00%
Missing dialogue indicators (quotation marks)
Target: ≤10% speech attributions without quotation marks
totalAttributions
10
unquotedAttributions
0
matches
(empty)
100.00%
Name drop frequency
Target: ≤1.0 per-name mentions per 100 words
totalMentions
38
wordCount
1316
uniqueNames
14
maxNameDensity
0.91
worstName
"Reyes"
maxWindowNameDensity
2
worstWindowName
"Reyes"
discoveredNames
Quinn
10
Army
1
Camden
2
TfL
1
GoPro
1
Reyes
12
London
2
Morris
1
Deptford
2
Superintendent
1
Federation
1
British
1
Museum
1
Harlow
2
persons
0
"Quinn"
1
"Reyes"
2
"Morris"
3
"Federation"
4
"Museum"
5
"Harlow"
places
0
"TfL"
1
"London"
2
"Deptford"
globalScore
1
windowScore
1
100.00%
Narrator intent-glossing
Target: ≤2% narration sentences with intent-glossing patterns
analyzedSentences
81
glossingSentenceCount
0
matches
(empty)
100.00%
"Not X but Y" pattern overuse
Target: ≤1 "not X but Y" per 1000 words
totalMatches
1
per1kWords
0.59
wordCount
1696
matches
0
"Not polished, but clean, the way shoes were"
100.00%
Overuse of "that" (subordinate clause padding)
Target: ≤2% sentences with "that" clauses
thatCount
1
totalSentences
140
matches
0
"knew that name"
100.00%
Paragraph length variance
Target: CV ≥0.5 for paragraph word counts
totalParagraphs
64
mean
26.5
std
23.44
cv
0.885
sampleLengths
0
9
1
71
2
20
3
7
4
7
5
11
6
49
7
74
8
15
9
46
10
7
11
18
12
78
13
3
14
7
15
44
16
5
17
42
18
7
19
20
20
8
21
70
22
10
23
6
24
1
25
2
26
32
27
47
28
30
29
6
30
6
31
12
32
97
33
37
34
27
35
47
36
12
37
20
38
48
39
16
40
56
41
4
42
3
43
27
44
7
45
55
46
32
47
51
48
1
49
24
98.94%
Passive voice overuse
Target: ≤2% passive sentences
passiveCount
2
totalSentences
111
matches
0
"been sealed"
1
"was tiled"
100.00%
Past progressive (was/were + -ing) overuse
Target: ≤2% past progressive verbs
pastProgressiveCount
0
totalVerbs
211
matches
(empty)
100.00%
Em-dash & semicolon overuse
Target: ≤2% sentences with em-dashes/semicolons
emDashCount
0
semicolonCount
1
flaggedSentences
1
totalSentences
140
ratio
0.007
matches
0
"She knew roughly where north was; the stairs had come down facing it."
100.00%
Purple prose (modifier overload)
Target: <4% adverbs, <2% -ly adverbs, no adj stacking
wordCount
1322
adjectiveStacks
0
stackExamples
(empty)
adverbCount
40
adverbRatio
0.030257186081694403
lyAdverbCount
9
lyAdverbRatio
0.00680786686838124
100.00%
Repeated phrase echo
Target: ≤20% sentences with echoes (window: 2)
totalSentences
140
echoCount
0
echoWords
(empty)
100.00%
Sentence length variance
Target: CV ≥0.4 for sentence word counts
totalSentences
140
mean
12.11
std
10.3
cv
0.851
sampleLengths
0
9
1
19
2
18
3
6
4
20
5
8
6
11
7
9
8
7
9
7
10
5
11
6
12
2
13
24
14
13
15
2
16
8
17
13
18
22
19
16
20
23
21
15
22
18
23
28
24
7
25
17
26
1
27
36
28
42
29
3
30
7
31
2
32
23
33
4
34
15
35
5
36
10
37
27
38
5
39
4
40
3
41
20
42
8
43
23
44
25
45
2
46
20
47
9
48
1
49
6
72.62%
Sentence opener variety
Target: ≥60% unique sentence openers
consecutiveRepeats
10
diversityRatio
0.4785714285714286
totalSentences
140
uniqueOpeners
67
70.92%
Adverb-first sentence starts
Target: ≥3% sentences starting with an adverb
adverbCount
2
totalSentences
94
matches
0
"Somewhere around sixty the air"
1
"Then another line, parallel, three"
ratio
0.021
58.30%
Pronoun-first sentence starts
Target: ≤30% sentences starting with a pronoun
pronounCount
38
totalSentences
94
matches
0
"It stopped smelling of Camden,"
1
"He looked about nineteen and"
2
"Her breath hung in front"
3
"She checked her watch, the"
4
"She filed the temperature away"
5
"He was younger than Quinn"
6
"She stopped two metres from"
7
"His face was calm."
8
"His eyes were open, and"
9
"He waved the coffee at"
10
"She stood and turned a"
11
"It lay thick across the"
12
"She could see her own"
13
"She watched him see it,"
14
"She pointed with her chin"
15
"He didn't have an answer"
16
"He drank some coffee instead."
17
"She crouched and touched one"
18
"She paced it."
19
"She put the memory back"
ratio
0.404
61.06%
Subject-first sentence starts
Target: ≤72% sentences starting with a subject
subjectCount
75
totalSentences
94
matches
0
"The stairs went down farther"
1
"Harlow Quinn counted them without"
2
"It stopped smelling of Camden,"
3
"That bothered her more than"
4
"He looked about nineteen and"
5
"Her breath hung in front"
6
"She checked her watch, the"
7
"She filed the temperature away"
8
"The platform opened in front"
9
"Arc lamps on tripods threw"
10
"The station had been sealed"
11
"Nobody had come down here"
12
"The dead man lay near"
13
"Callum Reyes stood over him"
14
"He was younger than Quinn"
15
"She stopped two metres from"
16
"The victim wore a charcoal"
17
"His face was calm."
18
"His eyes were open, and"
19
"Reyes followed her look and"
ratio
0.798
0.00%
Subordinate conjunction sentence starts
Target: ≥2% sentences starting with a subordinating conjunction
subConjCount
0
totalSentences
94
matches
(empty)
ratio
0
100.00%
Technical jargon density
Target: ≤6% sentences with technical-jargon patterns
analyzedSentences
52
technicalSentenceCount
2
matches
0
"The far wall was tiled like the rest, cream and oxblood, with the station's name picked out in a mosaic panel that had lost half its letters."
1
"Something else was nagging at her, and she crouched again and looked at the dead man's suit, at the inside breast pocket, which sat slightly open."
100.00%
Useless dialogue additions
Target: ≤5% dialogue tags with trailing filler fragments
totalTags
12
uselessAdditionCount
0
matches
(empty)
100.00%
Dialogue tag variety (said vs. fancy)
Target: ≤10% fancy dialogue tags
totalTags
4
fancyCount
2
fancyTags
0
"Quinn agreed (agree)"
1
"He looked (look)"
dialogueSentences
41
tagDensity
0.098
leniency
0.195
rawRatio
0.5
effectiveRatio
0.098
90.7913%