NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0MW299206N0T6ESW0K4EWX
Run Details
Model:
GPT-5.5
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
28.8s
Cost:
$0.034260
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.22%
Dialogue to Total Word Ratio
(details)
0.1092%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% words of dialogue between two thieves named Molly and Bill who are planning to break into a children's party to steal all the presents. Only include the prose of the scene without any markdown or commentary.
Model response
Messages
Meta
AI
“I’m telling you, Bill, it’s the perfect score.” “The perfect score? Molly, it’s a six-year-old’s birthday party.” “Exactly. Low security, high volume, distracted witnesses.” “Witnesses with sticky hands and unusually good memories.” Molly crouched behind the hedge across from the community hall, peering through a gap in the leaves. Balloons bobbed at the entrance. A banner read HAPPY BIRTHDAY, PIPPA! in glittery purple letters. Bill adjusted the fake mustache drooping over his upper lip. “Also, we are not stealing from children.” “We’re not stealing from children. We’re redistributing unopened merchandise from parents with disposable income.” “That’s stealing from children with extra steps.” “Fine. We’re liberating presents from the tyranny of wrapping paper.” Bill sighed. “You said this was an easy catering job.” “It is. We enter as entertainers.” “We are dressed as caterpillars.” “Very famous entertainers. The Wriggle Brothers.” “There are only two of us.” “Budget cuts.” A burst of shrieking laughter came from inside. Bill flinched. “Hear that? That’s not joy. That’s a security alarm made of sugar.” Molly pulled a folded napkin from her pocket and spread it on the grass. A crude map was drawn in crayon. “Here’s the layout. Cake table here. Bouncy castle here. Present mountain here.” “Why is there a dragon?” “That’s Uncle Steve. He’s wearing a dragon costume and, according to my research, has weak ankles.” “You researched Uncle Steve?” “I research everything.” “You spelled ‘presents’ with a z.” “It’s a tactical z.” Bill leaned closer. “What’s this red circle?” “Danger zone.” “The clown?” “No. The toddler with the recorder.” As if summoned, a piercing toot echoed from inside. Bill went pale. “We leave now.” “We do not leave. We adapt.” “Molly, I have burgled mansions, warehouses, and one surprisingly aggressive garden shed. I will not be taken down by a child playing Hot Cross Buns.” “You lack vision.” “I possess morals.” “You once stole a marble statue of a duck.” “It was insured.” “It was in a public fountain.” “It looked unhappy.” Molly folded the napkin. “Listen. We go in, perform a brief educational dance about sharing, roll the presents out on a trolley, and vanish before the pass-the-parcel finale.” “Educational dance?” “Improvised.” “I don’t improvise in antennae.” The hall door opened. A woman in a party hat poked her head out. “Are you the caterpillar magicians?” Molly sprang up. “Yes!” Bill whispered, “We are?” “Of course,” Molly whispered back. “Commit.” The woman smiled with the desperate brightness of an adult outnumbered by children. “Thank goodness. The puppet man cancelled after being bitten.” “By what?” Bill asked. The woman’s smile twitched. “No one is assigning blame.” Inside, the hall was chaos with streamers. Children thundered past like tiny, frosted meteors. On a long table, the presents gleamed in glossy paper and bows. Molly’s eyes shone. “Beautiful.” Bill elbowed her. “Don’t say beautiful at the evidence.” A small girl wearing a cardboard crown marched up to them. “Are you funny?” Molly bowed. “Extremely.” Bill gave a weak wiggle. “Moderately.” The girl narrowed her eyes. “Can you make presents disappear?” Molly froze. Bill said quickly, “No.” Molly said, “Theoretically.” The girl pointed at the table. “Mummy says nobody opens them until after cake. But Jamie keeps squeezing mine.” Bill looked toward the present pile. A boy in a pirate hat was indeed kneading a parcel like dough. Molly whispered, “Opportunity.” Bill whispered, “For decency.” He stepped forward, clapped his green mittens, and boomed, “Children! Who wants to learn the sacred caterpillar rule?” Thirty faces turned. Molly hissed, “What are you doing?” “Improv.” Bill raised one finger. “Rule one: don’t touch presents that aren’t yours.” A chorus of groans erupted. “Rule two,” Molly added, recovering, “all unattended presents must be moved to a secure location by qualified caterpillar professionals.” Bill glared. A boy shouted, “Can we watch?” “No,” Molly said. “Yes,” Bill said. “Because the secure location is right here, where everyone can see them.” Molly’s smile stiffened. “Bill.” “Also,” he continued, “any present-squeezers must help stack chairs.” Jamie the pirate released the parcel. The birthday girl tugged Molly’s sleeve. “If you’re magicians, make Uncle Steve disappear.” Across the room, the dragon-costumed man was stuck halfway into the bouncy castle entrance, roaring for assistance. Molly stared at the present table, then at Bill, then at Uncle Steve. Bill folded his arms. “Well?” Molly sighed. “Fine. We rescue the dragon, protect the loot, eat cake, and leave empty-handed.” “That’s the spirit.” “But I’m taking a party bag.” “One party bag.” “And a balloon.” “One balloon.” “And if there’s a duck-shaped gift—” “Molly.” “Fine. No ducks.”
created_at
"2026-04-24T21:03:30.118Z"
4/24/2026, 9:03:30 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
626
moderation_latency
(null)
generation_time
28641
tokens_prompt
77
tokens_completion
1506
native_tokens_prompt
66
native_tokens_completion
1131
native_tokens_completion_images
0
native_tokens_reasoning
0
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
"completed"
usage
0.03426
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_013b01d506ce9e750169ebdaa236d48194811bfe6df1503fa2"
is_byok
false
latency
305
model_permaslug
"openai/gpt-5.5-20260423"
provider_name
"OpenAI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777064610-KomeSny2N6UMxEtpQ7Kn"
session_id
(null)
api_type
"completions"
id
"gen-1777064610-fO0Xvh6RVVdNLLfUV2aX"
upstream_id
"resp_013b01d506ce9e750169ebdaa236d48194811bfe6df1503fa2"
total_cost
0.03426
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
759
0.22%
Dialogue to Total Word Ratio
Ratio: 58.93%, Deviation: 11.07%
neededClean
false
wordsTotal
767
wordsDialogue
452
0.1092%