NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX41FAT58N2V2N1G738VNKPJ
Run Details
Model:
GPT-5.6 Sol
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
15.4s
Cost:
$0.021240
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
9.61%
Matches word count
(details)
99.51%
Dialogue to Total Word Ratio
(details)
54.5595%
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 spread the blueprint across the café table, pinning its corners with four doughnuts. Bill immediately ate the northeast exit. “That was our escape route,” Molly said. “It was custard.” “It was an architectural document.” “It had sprinkles.” Outside, rain polished the windows. Inside, the lunch crowd ignored them, largely because Molly had labeled the blueprint CHILDREN’S PARTY HEIST in pink crayon, which made it seem less criminal and more like homework. “We enter at fourteen hundred hours,” she said. “Disguised as entertainers.” “What kind?” “Magicians.” “I don’t know magic.” “Neither do children. They think sleeves are supernatural.” Bill frowned at the drawing. “Why are we stealing birthday presents?” “Because Lady Penelope Crumb’s twins receive extravagant gifts. Gold trains. Diamond rattles. Possibly a pony.” “You can’t wrap a pony.” “With enough paper, anything is possible.” Molly tapped the ballroom. Her fingernail left a dot of raspberry jam beside the bouncy castle. The mansion’s party occupied the east wing, with guards at every door and sugar-fueled children everywhere. “The presents are stacked here,” she said. “We create a diversion, load them into laundry carts, and vanish.” “What diversion?” “You perform.” “Perform what?” “The Disappearing Ferret.” “We don’t own a ferret.” “We’ll use a sock.” “A sock isn’t convincing.” “Neither are you, but I still bring you.” Bill considered this insult while buttering the remaining western staircase. Across the café, a waitress glanced over. Molly smiled and casually covered the blueprint with a menu. Bill whispered, loudly, “What about the children?” “We don’t steal them.” “I meant witnesses.” “We distract them with cake.” “They’ll already have cake.” “More cake.” “Diabolical.” The waitress approached. Molly ordered tea. Bill ordered three emergency éclairs, claiming planning consumed glucose. When she left, Molly revealed two invitations stolen from Lady Crumb’s secretary. “Names?” Bill asked. “Mystico Molly and William the Astonishing.” “Can I be William the Terrifying?” “At a children’s party?” “William the Mildly Concerning?” “You may be Bill.” He sighed. The rain intensified, drumming like impatient fingers. Molly outlined the route: service entrance, pantry, ballroom, gift mountain, laundry chute. Bill’s expression grew increasingly troubled. “The chute goes to the basement,” he said. “Correct.” “And the laundry carts?” “Also basement.” “And the pony?” “Contingency.” “Meaning?” “We ride it out.” Bill stared at her. “Through the kitchen?” “Preferably through the gates.” “Do you know how to ride?” “No.” “Does the pony?” “Probably.” The waitress returned with tea and éclairs. She placed the plate down, then pointed at the exposed blueprint. “My son’s attending that party,” she said. Molly froze. Bill swallowed an éclair whole. “How lovely,” Molly managed. “He hates magicians.” “Even better,” said Bill. “Low expectations.” The waitress leaned closer. “The presents aren’t in the ballroom. They’re locked upstairs.” “Why tell us?” Molly asked. “Because Lady Crumb never tips.” She drew a staircase on the plan, took one doughnut, and smiled. “Bring me the pony.”
created_at
"2026-07-09T18:15:38.572Z"
7/9/2026, 6:15:38 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
674
moderation_latency
(null)
generation_time
15349
tokens_prompt
77
tokens_completion
1013
native_tokens_prompt
66
native_tokens_completion
697
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"
service_tier
"default"
usage
0.02124
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_0e7d49aa5ab1cbda016a4fe54aa268819484ccff798505536e"
is_byok
false
latency
217
model_permaslug
"openai/gpt-5.6-sol-20260709"
provider_name
"OpenAI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1783620938-WRCuD2kJasLWeqtvk6TJ"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783620938-QY2EM0W0jqx8ENIiHcDk"
upstream_id
"resp_0e7d49aa5ab1cbda016a4fe54aa268819484ccff798505536e"
total_cost
0.02124
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
9.61%
Matches word count
n/a
neededClean
false
words
478
99.51%
Dialogue to Total Word Ratio
Ratio: 50.31%, Deviation: 0.31%
neededClean
false
wordsTotal
479
wordsDialogue
241
54.5595%