NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0MZR32F5QAV34NFQ6VY9AP
Run Details
Model:
GPT-5.5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
30.9s
Cost:
$0.032790
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
78.63%
Dialogue to Total Word Ratio
(details)
39.3162%
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 unrolled the crayon-smeared floor plan across the bonnet of Bill’s van and pinned the corners with four stolen cupcakes. The drawing showed a church hall, a bouncy castle, a trestle table shaped by her imagination into a fortress, and forty-seven presents stacked beneath a banner that read HAPPY 6TH BIRTHDAY, OLIVER! in glitter glue. Bill squinted at it. His balaclava was on inside out, which made him look like a confused sock puppet. “That’s not a floor plan,” he said. “That’s a refrigerator drawing.” “It’s intelligence,” said Molly. “Acquired from a reliable source.” “You mugged a six-year-old?” “I babysat his cousin. Briefly. Until the incident with the hamster.” Bill looked at the hall through binoculars. A clown was unloading balloon animals from a hatchback. Two mothers carried trays of sausage rolls. A man in a dinosaur costume stood smoking behind the bins, looking as if evolution had disappointed him. “Security is heavier than expected,” Bill said. “Parents,” Molly muttered. “The most dangerous mammals in any suburban ecosystem.” They crouched behind the van while a troop of children wearing paper crowns streamed past, shrieking like tiny police sirens. Molly checked her watch. “We go in at fourteen hundred hours, during Pass the Parcel. Maximum distraction. Minimum dignity.” “What’s our cover?” “Entertainment.” Bill’s face collapsed. “No.” “Yes.” “I told you after the retirement home job, I’m not doing magic again.” “You made Mrs. Dobson’s teeth disappear.” “They were in my pocket for sentimental reasons.” Molly opened the van’s side door. Inside hung two costumes: a fairy princess gown with tactical pockets sewn into the skirt, and a pirate outfit with a plastic cutlass and suspiciously large boots. Bill stared at them as though they were prison uniforms. “I’m not wearing the pirate hat,” he said. “You are absolutely wearing the pirate hat. Children respect authority when it has a skull on it.” “They’ll know I’m not a real pirate.” “Bill, no one thinks you’re a real anything.” He sighed and put on the hat. It drooped over one eye, giving him the air of a man being slowly eaten by felt. Molly slipped into the fairy dress, strapped a gift sack beneath her wings, and produced a clipboard. “Phase one: entry. Phase two: dazzle. Phase three: extract presents via rear corridor while everyone is hypnotized by your sword dance.” “My what?” “Improvised sword dance.” “I don’t improvise below the waist.” “You will today.” Across the car park, the birthday boy appeared at the hall door clutching a toy rocket. He wore a badge the size of a saucer that said I AM OLIVER. He examined the arriving guests with the grave suspicion of a customs officer. Bill softened. “He’s quite small.” “They usually are at six.” “And those presents are probably mostly socks, puzzles, books…” “High resale value on unopened unicorn slime,” Molly said, but her voice had lost some bite. A little girl tripped nearby, scattering a pile of wrapped gifts onto the pavement. Molly moved without thinking, catching the biggest box before it hit a puddle. The girl stared up at her. “Are you the fairy?” Molly hesitated. “Technically.” “Can you fix my present? The bow fell off.” Bill watched as Molly knelt, retied the ribbon with expert fingers, and added a flourish that made it look shop-bought. The girl beamed and ran inside. Bill folded his arms. “Master criminal.” “Shut up.” The clown began honking a horn. Inside, children cheered. Someone shouted that the magician was late. Molly looked at the mountain of presents, then at Bill’s sagging pirate hat, then at the hall full of sugar-fuelled witnesses. “Revised plan,” she said. “Oh good. Does it involve fleeing?” “It involves cake.” They entered through the front doors to thunderous applause from people who assumed they had been booked. Bill was handed a bubble wand. Molly was given a glitter microphone. For forty minutes, they twisted balloons, led a treasure hunt, and survived a barrage of questions about whether fairies paid taxes. When the candles were lit, Oliver made a wish so fiercely his cheeks puffed like bellows. Molly caught Bill eyeing the presents. “Don’t,” she whispered. “I wasn’t.” “You were.” “I was wondering if we should steal the clown’s car instead.” Molly smiled despite herself. “After cake.” By the time they left, their sack was full, but only with leftover sandwiches, party hats, and one thank-you card from Oliver that read BEST FAIRY AND PIRIT EVER. Bill tucked it carefully into his coat. “Terrible haul,” Molly said. Bill nodded. “Catastrophic.” From inside came another burst of laughter. Molly glanced back once. “Same time next year?” Bill adjusted his pirate hat with dignity. “Only if I get to be the fairy.”
created_at
"2026-04-24T21:05:30.739Z"
4/24/2026, 9:05:30 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3908
moderation_latency
(null)
generation_time
28994
tokens_prompt
77
tokens_completion
1485
native_tokens_prompt
66
native_tokens_completion
1082
native_tokens_completion_images
0
native_tokens_reasoning
26
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
0.03279
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_04ddf958e270f9b80169ebdb1ca81481969d258f2866d76664"
is_byok
false
latency
435
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-1777064732-Z3aDF9frhoeHoW0OktBV"
session_id
(null)
api_type
"completions"
id
"gen-1777064732-59JNyhITBQLFJ2uly1yP"
upstream_id
"resp_04ddf958e270f9b80169ebdb1ca81481969d258f2866d76664"
total_cost
0.03279
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
782
78.63%
Dialogue to Total Word Ratio
Ratio: 32.19%, Deviation: 2.19%
neededClean
false
wordsTotal
789
wordsDialogue
254
39.3162%