NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0NBMMDG6GV76AZ9P7WGMGC
Run Details
Model:
GPT-5.5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
28.4s
Cost:
$0.034560
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.33%
Dialogue to Total Word Ratio
(details)
0.1668%
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 spread the birthday invitation across the café table as if it were a map to buried treasure. In a sense, it was. The treasure had glitter stickers, cartoon dinosaurs, and the words Presents may be left on the gift table beside the bouncy castle printed in cheerful blue ink. Bill leaned over it, squinting. A smear of jam glistened on his chin. He had insisted that no criminal enterprise could be planned without scones. “A bouncy castle,” Molly said. “That’s our complication.” “That’s our opportunity,” Bill replied. “It is full of children.” “Exactly. No one expects professionalism near a bouncy castle.” Molly tapped the invitation with one black-painted fingernail. “We’re not stealing from children, technically. We’re stealing from a party.” Bill nodded solemnly. “A party is an institution. Institutions can absorb loss.” “It says here the birthday boy is turning six.” “A small institution.” Molly gave him a look. Bill dabbed his chin with a napkin. “Fine. A medium-small institution with sticky fingers and no tax status.” Outside the café window, rain slid down the glass like nervous sweat. Molly preferred weather that understood crime. Bill preferred weather that allowed hats. Today they had both. Their target was the annual Maple Street birthday extravaganza, hosted by the Harringtons, a family famous for hiring magicians who used live doves and for giving party bags heavy enough to sprain a wrist. Last year, according to Molly’s research, each child had received a miniature telescope, a chocolate fountain voucher, and a kazoo plated in something suspiciously shiny. This year, the gift table would be enormous. Molly removed two crayons from her coat pocket and placed them beside the invitation. Bill frowned. “Why have you got crayons?” “To blend in.” “You’re thirty-four.” “And youthful.” “You once complained that a trampoline had poor lumbar support.” Molly ignored this and drew a rectangle. “Garden. Here’s the bouncy castle. Here’s the cake station. Here’s the gift table.” Bill pointed at a blob near the gate. “What’s that?” “That is you.” “I’m a blob?” “You’re undercover as entertainment.” Bill sat straighter. “I will not be a clown.” “You own three waistcoats and a collapsing top hat.” “That is formal wear.” “You know balloon animals.” “I know balloon sausage. Everything becomes sausage.” Molly smiled. “Perfect. Children have low standards and high volume.” Bill looked wounded, but only briefly, because the waitress arrived with another scone. He took it as a sign from fate. Their plan, if dignified with the word, involved distraction, misdirection, and a large laundry hamper painted to resemble a robot. Molly would enter as the robot’s “operator,” claiming it was an educational demonstration about recycling. Bill would create balloon sausages near the paddling pool until every adult was too baffled to intervene. Then Molly would roll the hamper past the gift table, open the robot’s stomach, and scoop. “No actual running,” Molly said. “Running implies guilt.” “Waddling?” “Purposeful waddling.” Bill considered this. “What if someone asks why the recycling robot is eating presents?” “It’s teaching consumer responsibility.” “Brilliant.” “What if someone asks why the magician has only made balloon sausages?” Bill lifted his chin. “Artistic signature.” Molly folded the invitation. “We leave through the side alley, transfer the presents to the pram, and vanish.” “We have a pram?” “You have a pram.” “I have a shopping trolley with a blanket.” “Criminals improvise.” For a moment, they sat in satisfied silence, listening to teaspoons clink and rain patter. It was almost beautiful, except for the moral ugliness and Bill chewing with his mouth open. Then Molly’s phone buzzed. She checked it, and her expression changed. “What?” Bill asked. She turned the screen toward him. A message from her sister: Don’t forget Saturday! Freddie’s party at Maple Street. Bring a gift. Also, you promised to help supervise the bouncy castle. Bill blinked. Molly blinked back. “The birthday boy,” Bill said slowly, “is your nephew?” “Apparently.” “You didn’t know your own nephew’s age?” “He grows unpredictably.” Bill looked at the invitation, then at the crayons, then at the scone. “So the heist is off?” Molly sighed the sigh of a woman watching a fortune in wrapping paper dissolve. “The heist is revised.” “To what?” She picked up a crayon and drew a new plan over the old one. “We break into the party…” “Yes?” “We steal all the presents…” “Yes?” “And hide them in the garage until cake, so the children don’t open them early and start a riot.” Bill stared at her. “It’s still a heist,” she said. “Just benevolent.” He brightened. “Can I still be the magician?” “You may still be the magician.” “And the robot?” “The robot is now educational.” Bill grinned. “Molly, I think we’ve gone straight.” “Don’t be vulgar,” she said. “We’ve gone seasonal.”
created_at
"2026-04-24T21:12:00.404Z"
4/24/2026, 9:12:00 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4610
moderation_latency
(null)
generation_time
28265
tokens_prompt
77
tokens_completion
1576
native_tokens_prompt
66
native_tokens_completion
1141
native_tokens_completion_images
0
native_tokens_reasoning
55
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
0.03456
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_04ad75fb177180900169ebdca07c0c819586f6c8540d689517"
is_byok
false
latency
318
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-1777065120-Nxp7eoOCW3hfkQ5q09sB"
session_id
(null)
api_type
"completions"
id
"gen-1777065120-bAOmyX4CiZu9szL8H84F"
upstream_id
"resp_04ad75fb177180900169ebdca07c0c819586f6c8540d689517"
total_cost
0.03456
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
791
0.33%
Dialogue to Total Word Ratio
Ratio: 40.68%, Deviation: 10.68%
neededClean
false
wordsTotal
794
wordsDialogue
323
0.1668%