NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0P6PTSPYAXH7AMMVBRRVS8
Run Details
Model:
GPT-5.5
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
29.1s
Cost:
$0.033930
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
63.71%
Dialogue to Total Word Ratio
(details)
31.8564%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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
Molly crouched behind the azalea hedge, peering through a pair of opera glasses she had stolen from a man who had stolen them first. Across the lawn, a garden seethed with balloons, bunting, and children sticky with cake. “Target acquired,” Molly whispered. Bill adjusted his fake moustache, which was slipping toward his chin. “You mean the bouncy castle?” “The presents, Bill.” “Right. Obviously. I was assessing escape routes.” “The bouncy castle is not an escape route.” “It could be, with imagination.” Molly lowered the glasses. “There are forty-three wrapped parcels on that table. Ribbons. Tags. One suspiciously large box.” “Bicycle?” “Probably magician’s assistant.” Bill squinted. “Do children still appreciate magicians?” “Children appreciate chaos. That’s why we blend in.” He looked down at his outfit: enormous polka-dot trousers, yellow braces, and shoes the size of rowing boats. “I feel conspicuous.” “You’re dressed as a clown at a party with three clowns.” “Yes, but I’m the only one sweating morally.” A shriek erupted as a small boy struck a piñata with the grave determination of a medieval executioner. Molly consulted her notebook. “Phase one: enter through side gate during piñata frenzy.” “Phase two?” “You distract the adults.” “How?” “Make balloon animals.” Bill held up limp balloons. “I can make a worm.” “Excellent. Make a zoo of worms.” “And you?” “I wheel the presents away in that decorated pram.” Bill followed her gaze to a pram covered in streamers and glitter. “Is there a baby in it?” “Not after I check.” “Molly.” “What?” “We are not stealing from babies.” “We’re not stealing from babies. We’re stealing near babies.” “That feels like a legal distinction, not a moral one.” She sighed. “Fine. I check thoroughly, then steal the empty pram.” A woman in a fairy costume appeared with a tray of sausage rolls. Bill inhaled deeply. “New complication,” he said. “Catering.” “No eating on the job.” “It builds credibility. Clowns eat.” “Clowns terrify.” “Then I’m already credible.” Molly tucked a sack beneath her cardigan. “Remember, no improvising.” “I never improvise.” “You once escaped a jeweler’s by pretending to be a visiting duke.” “That was scripted by panic.” “And you knighted a police horse.” “It deserved recognition.” The piñata burst. Sweets rained down. Children dived like stockbrokers during a market crash. Molly snapped her fingers. “Move.” They shuffled through the side gate. Bill immediately stepped on his own shoe and lurched into a trellis. “Natural clowning,” he muttered. “Convincing,” Molly said. “Painfully.” A parent waved. “You must be the entertainment!” Bill bowed. His moustache fell off. A toddler picked it up and screamed. “Behold,” Bill announced, “the vanishing moustache!” The toddler stopped screaming and ate it. Molly slipped toward the gift table, smiling with the serene menace of a shark in sensible shoes. Each present bore a bright label: For Daisy. Happy Birthday! Love Grandma. She paused at a box wrapped in silver stars. Bill, meanwhile, inflated a balloon with heroic effort. It squeaked into a long, sad tube. “What animal is that?” asked a girl. “An endangered baguette,” Bill said. Children gathered. “Make a giraffe!” “Make a dragon!” “Make a tax accountant!” Bill glanced at Molly, eyes wide. “Molly?” She had one hand on a present, but Daisy, the birthday girl, stood beside the table, wearing a paper crown and solemnly hugging a battered stuffed rabbit. “Are you a clown too?” Daisy asked. Molly looked at the presents, then at Daisy. “Technically, I’m procurement.” “What’s that?” “Someone who finds things.” Daisy’s lip trembled. “I lost Mr. Nibbles’s hat.” Molly froze. “The rabbit has a hat?” “He did.” Bill called, “Molly, the children demand a tax accountant!” Molly scanned the grass, spotted a tiny felt top hat beneath a chair, and picked it up. Daisy gasped as if Molly had produced sunlight. “You found it!” “Yes,” Molly said. “That is my curse.” Daisy hugged her knees. “You’re the best clown.” Bill arrived, covered in balloons and shame. “We have a problem.” “I know,” Molly said. “Is it conscience?” “Unfortunately.” He nodded gravely. “I hate when that happens.” Molly placed the silver present back on the table. “Abort heist.” “Thank cake?” “Thank cake.” Later, behind the hedge, they ate stolen sausage rolls. “We failed,” Bill said. Molly licked icing from her thumb. “We performed a service.” “We recovered a hat.” “And liberated six sausage rolls from certain neglect.” Bill raised one in salute. “To crime, narrowly avoided.” Molly clinked hers against it. “To better targets.” From the garden came Daisy’s voice. “Goodbye, best clowns!” Bill smiled. “We should rob a bank next.” Molly nodded. “At least banks don’t have rabbits.”
created_at
"2026-04-24T21:26:47.393Z"
4/24/2026, 9:26:47 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1552
moderation_latency
(null)
generation_time
29036
tokens_prompt
77
tokens_completion
1491
native_tokens_prompt
66
native_tokens_completion
1120
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
0.03393
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_08fe003b35e190950169ebe01777148196baadbfc87842b1b3"
is_byok
false
latency
215
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-1777066007-CqCZcVG73U6kTY81YfNe"
session_id
(null)
api_type
"completions"
id
"gen-1777066007-s18NiZsO5zzV7noLCiNJ"
upstream_id
"resp_08fe003b35e190950169ebe01777148196baadbfc87842b1b3"
total_cost
0.03393
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
764
63.71%
Dialogue to Total Word Ratio
Ratio: 47.00%, Deviation: 3.00%
neededClean
false
wordsTotal
766
wordsDialogue
360
31.8564%