NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0NMSJTKFQ4RX18BXG9ZBEJ
Run Details
Model:
GPT-5.5
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
30.5s
Cost:
$0.035220
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
95.72%
Dialogue to Total Word Ratio
(details)
47.8604%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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 unfolded the crayon map on the dashboard, smoothing it over a smear of powdered doughnut. The map had been drawn by Bill’s nephew, allegedly “from memory,” which explained why the house was shaped like a pirate ship and the driveway was labeled THE PLACE WHERE DAD CRIED. Across the street, balloons bobbed from the mailbox of number 14. A banner sagged between two porch columns: HAPPY 7TH BIRTHDAY, CHLOE! Beneath it, a mountain of gift bags glittered in the front room window like treasure from a very polite dragon’s cave. Bill adjusted his fake mustache. It was purple. Molly had told him not to buy disguises from a vending machine at the bowling alley, but Bill respected bargains more than warnings. “We go in as entertainers,” Molly said. Bill looked down at his outfit: black turtleneck, black trousers, black gloves, and a cape covered in sequins. “I thought we were burglars with flair.” “You look like a magician who was fired by a funeral home.” “That is a strong brand.” “No, Bill. The plan is simple. We enter through the side gate, say we’re the backup balloon artists, locate the presents, and move them to the van.” Bill squinted at the house. “What if they ask us to make balloons?” Molly produced a long red balloon and twisted it expertly into a sword. Bill stared. “When did you learn that?” “Prison workshop.” “You were in prison?” “Community center. But the instructor had energy.” A shriek of laughter came from the backyard. Children stampeded past a window wearing paper crowns. One had cake frosting in both eyebrows and the hollow-eyed determination of a tiny Viking. Bill swallowed. “There are so many of them.” “They’re seven. Their bones are mostly pudding.” “That makes them unpredictable.” Molly tapped the map. “Presents are here, by the window. We need a distraction.” Bill brightened. “I can juggle.” “You can’t juggle.” “I can drop three oranges in a rhythm.” “That’s called littering.” Before Bill could defend his art, the front door opened and a woman in a unicorn headband stepped onto the porch, carrying a clipboard. Molly ducked. Bill ducked too late, headbutting the horn. “Ow,” he whispered. “The unicorn sees all.” The woman scanned the street, then shouted, “Are you the balloon people?” Molly popped up with a grin wide enough to be legal tender. “Indeed we are!” Bill followed, clutching his forehead. “And I am Professor Inflato.” “No,” Molly hissed. “Doctor Inflato?” The woman waved them in. “You’re late. The magician canceled.” Molly smiled harder. “Tragic.” Inside, the house smelled of vanilla icing, carpet cleaner, and panic. Gift bags covered the dining table. Boxes leaned in glossy stacks. Molly’s fingers twitched with professional appreciation. Bill whispered, “That one has a robot on it.” “We are not shopping.” “I’m not. I’m identifying assets.” A small girl in a tiara appeared before them. “Can you make a giraffe?” Molly snapped a balloon from her belt. “Absolutely.” Bill leaned close. “Can you?” “No.” Three minutes later, Molly handed the girl something resembling a dachshund that had made several poor life choices. The girl gasped. “It’s a dragon!” “Yes,” Molly said. “A rare flat-necked dragon.” Soon children swarmed them, chanting for swords, crowns, and one emotionally specific request for “a snake who is also my dad.” Molly twisted balloons at heroic speed while Bill backed toward the gift table, opening his cape like a sparkly curtain. “Stage one,” he murmured. “Commencing.” A boy with frosting eyebrows pointed. “Why is that man stealing Chloe’s presents?” Bill froze with a gift bag halfway under his cape. Molly spun around. “He’s not stealing. He’s preparing a magic trick.” The children went silent. Bill whispered, “Molly.” “Make them disappear,” she said through her teeth. Bill lifted the bag, flung the cape over it, and shouted, “Behold!” Nothing happened. A music box inside the bag began playing “Pop Goes the Weasel.” The children leaned forward. Bill panicked, kicked the table, and twelve presents slid into his cape. He staggered backward, tripped over a plush unicorn, and crashed into the cake table. Frosting exploded across his black outfit. The purple mustache migrated to his cheek. For one stunned second, nobody moved. Then Chloe clapped. “Best magician ever!” The room erupted. Children cheered, parents filmed, and Bill, lying in buttercream, looked deeply betrayed by success. Molly bowed. “Thank you. For our next trick, we will move all gifts to the backyard for inspection.” The unicorn-headband woman frowned. “Inspection?” “Safety,” Molly said. “Batteries. Choking hazards. Cursed dolls.” At that exact moment, Bill sneezed and a toy robot fell out of his cape, marched three steps, and announced, “INTRUDER ALERT.” The room went quiet again. Molly picked up a balloon sword. Bill slowly stood, dripping icing. “New plan?” he asked. Molly nodded toward the door as frosting-eyebrow boy grinned like a tiny detective. “Run,” she said. They ran empty-handed, chased by thirty children wielding balloon dragons, while behind them Chloe shouted, “Come back next year!”
created_at
"2026-04-24T21:17:00.385Z"
4/24/2026, 9:17:00 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
788
moderation_latency
(null)
generation_time
30443
tokens_prompt
77
tokens_completion
1487
native_tokens_prompt
66
native_tokens_completion
1163
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
0.03522
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_0a24d547e0f529910169ebddcc8138819498648a0b78d50858"
is_byok
false
latency
420
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-1777065420-Y5x3Z98S9SCBnHKNoSNi"
session_id
(null)
api_type
"completions"
id
"gen-1777065420-jJcv1kLPIhaZOpmsykTo"
upstream_id
"resp_0a24d547e0f529910169ebddcc8138819498648a0b78d50858"
total_cost
0.03522
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
829
95.72%
Dialogue to Total Word Ratio
Ratio: 30.94%, Deviation: 0.94%
neededClean
false
wordsTotal
834
wordsDialogue
258
47.8604%