NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHK9CRNFNVGVCTJCD4NT3HR2
Run Details
Model:
GPT-5.1
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
1m 41s
Cost:
$0.047883
Provider:
OpenAI
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
The getaway van smelled like frosting and bad decisions, which, to be fair, was mostly Bill's cologne. He'd parked two blocks from the suburban house, where a helium balloon arch swayed like an overeager guard dog. Molly sat in the passenger seat, knees knocking against a cardboard box labeled "Party Supplies" in handwriting that looked suspiciously like ransom-note practice. "Tell me again why we're robbing a seven-year-old," Molly whispered. "Because kids' parties are the Fort Knox of unguarded consumer goods," Bill said. He flicked open a spiral notebook, blueprints of the backyard printed in pastel crayon. "Last time we hit a retirement home bingo night, you cried during cake," Molly said. "Those people had stories, Molly; kids just have stickers and a sugar problem," Bill replied. Outside, shrieks of tiny humans drifted over, punctuated by the wet thump of something hitting a piñata. A banner reading "HAPPY 7TH, OLIVIA!" sagged across the fence like it was already tired of this timeline. "I'm not wearing a clown nose again; it left a ring on my soul," Molly muttered. "It's a unicorn theme; you'll blend in as 'Disillusioned Aunt From Out Of Town,'" Bill said. "Your code names get longer every job," Molly said. "Length confuses witnesses; by the time they finish saying it, we're gone," Bill said. "So what's the plan, O Criminal Einstein of Cupcake Lane?" Molly asked. "We infiltrate as party entertainment, locate the present table, swap in decoys, exit with loot," Bill said. Bill had circled the present table three times on his crude map, then added a little halo above it. To him, loot was sacred; to Molly, loot was just another way to pay rent without talking to landlords. A faint chorus of "Baby Shark" leaked from the house, the universal anthem of parental surrender. Molly shuddered as though someone had walked over her grave and left plastic confetti. "Decoys meaning those educational wooden toys nobody wants?" Molly said. "Exactly; we're redistributing joy from phonics to laser-tag robots," Bill said. "You're aware we're villains in a children's morality tale, right?" Molly said. "Villains get all the good lines and none of the nap schedules," Bill replied. "Fine, but I'm not doing balloon animals; last time I accidentally made a noose," Molly said. "Relax; you just twist until it looks vaguely festive and not legally actionable," Bill said. "What about security, besides a feral piñata and three overcaffeinated dads?" Molly asked. "Intel says one magician, one face-painter, zero threat; worst case, we bribe with glitter," Bill said. He tapped the drawing of the gift pile, a colorful rectangle fortress beside the bouncy castle. In the margin he'd written "Objective Alpha" and drawn a smiley face wearing a ski mask. A gust of wind rattled the van, carrying with it the scent of grilled hotdogs and impending lawsuits. Molly adjusted her sequined blazer, which had seen better cons and worse lighting. Somewhere inside, a child wailed; the party had begun. "If glitter's involved, this is officially war," Molly said.
created_at
"2026-02-16T13:15:58.961024+00:00"
2/16/2026, 1:15:58 PM
model
"openai/gpt-5.1-20251113"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
9573
moderation_latency
(null)
generation_time
100507
tokens_prompt
77
tokens_completion
2088
native_tokens_prompt
66
native_tokens_completion
4780
native_tokens_completion_images
0
native_tokens_reasoning
4073
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"completed"
usage
0.0478825
router
(null)
provider_responses
0
id
"resp_073004e8f323338d0169931829bcf481978734814c7da32391"
status
200
is_byok
false
latency
190
endpoint_id
"764eb97f-8bab-4326-b29b-7a8799b00a70"
provider_name
"OpenAI"
model_permaslug
"openai/gpt-5.1-20251113"
api_type
"completions"
id
"gen-1771247657-XRzvzSRojmlWeZ4xX8jN"
upstream_id
"resp_073004e8f323338d0169931829bcf481978734814c7da32391"
total_cost
0.0478825
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
98.7124%
Matches word count
n/a
neededClean
false
words
506
29.5073%
Dialogue to Total Word Ratio
Ratio: 45.06%, Deviation: 4.94%
neededClean
false
wordsTotal
506
wordsDialogue
228
64.1098%