NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX40WBF2Z53KYTH628MZZMF7
Run Details
Model:
GPT-5.6 Sol
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
17.1s
Cost:
$0.022680
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
33.97%
Dialogue to Total Word Ratio
(details)
16.9870%
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 crayon map across the café table and pinned its corners with sugar bowls. A purple dragon occupied the ballroom. A green blob marked the kitchen. Thirty stick figures surrounded a towering birthday cake. Bill frowned. “Why does the dragon have spectacles?” “That’s the magician.” “He looks dangerous.” “He makes balloon dachshunds.” “Exactly. Anyone who can twist sausages without remorse is dangerous.” Rain scribbled against the windows. Around them, spoons clinked and the espresso machine hissed like an accomplice with asthma. Molly tapped a red X beside the drawing of a mountain of presents. “We enter at fourteen hundred hours.” “You mean two o’clock.” “I mean fourteen hundred hours. Crime sounds unprofessional before lunch.” “We’re robbing an eight-year-old.” “We’re stealing presents from forty overprivileged children. It’s practically taxation.” “Tax collectors don’t hide inside pantomime horses.” “The good ones do.” Bill examined the costume slumped beneath the table. Its papier-mâché head wore a permanently astonished expression, which matched his own. “Which end am I?” “The back.” “I was the back last time.” “You’re taller.” “That makes no anatomical sense.” “It makes theatrical sense. Audiences enjoy a horse with a gradual slope.” Molly produced two laminated invitations. They featured glitter, rockets, and the words TOMMY’S SUPER SPACE BIRTHDAY in explosive silver lettering. “Where did you get those?” “I made them.” “They say we’re Trevor and Mildred Horse.” “We needed cover identities.” “Horses don’t have surnames.” “Tell that to Sarah Jessica Parker.” Bill lowered his voice as a waitress passed. Molly smiled innocently and colored the magician’s dragon hat blue. “What about security?” “One grandmother at the entrance.” “Armed?” “With a clipboard.” “Those are worse. Clipboards create authority.” “You’ll distract her.” “How?” “Neigh.” “I have range, Molly. Give me motivation.” “You’re a horse desperate to use the lavatory.” Bill considered this with professional seriousness. Outside, a bus sprayed gutter water over a cyclist, providing the afternoon’s second-most elaborate disguise. Once inside, Molly explained, they would perform a short routine, retreat through the dining room, load the presents into the costume’s hollow body, and trot out before cake. Bill counted the parcels drawn around the X. “Forty presents won’t fit.” “They will if we remove the horse’s ribs.” “I’m beginning to understand why it looks surprised.” “Focus. The largest boxes go in the torso. Gift cards in the tail.” “And if a child pulls it?” “Kick gently.” “You cannot gently kick a child.” “Then whinny sternly.” Bill leaned back. “What if they ask us to dance?” “We gallop.” “What if the magician suspects us?” “We expose him.” “For what?” Molly pointed at the dragon. “That hat is evidence enough.” The waitress deposited their bill. Molly slid it toward Bill. “You pay.” “Why me?” “You’re the rear investor.” Bill sighed, opened his wallet, and stared at the costume. “One condition.” “What?” “I control the legs.” “Fine.” “And afterwards, we donate half.” “To charity?” “To my chiropractor.” Molly folded the map. “Deal. Meet outside the party at thirteen forty-five.” “Quarter to two.” She glared. “Sorry. Thirteen forty-five.” Together they rose, carrying the horse between them. Its head struck a hanging lamp, spun around, and stared accusingly at Bill. He whispered, “This job already has a witness.”
created_at
"2026-07-09T18:05:16.651Z"
7/9/2026, 6:05:16 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
969
moderation_latency
(null)
generation_time
16992
tokens_prompt
77
tokens_completion
1099
native_tokens_prompt
66
native_tokens_completion
745
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
0.02268
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_004718665a3f228c016a4fe2dd09c88190a3cf138ec3cb18ec"
is_byok
false
latency
523
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-1783620316-RPQ6Bflee7rmkFJTDRRP"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783620316-V9sVMn2QMY3XWGDXJyIG"
upstream_id
"resp_004718665a3f228c016a4fe2dd09c88190a3cf138ec3cb18ec"
total_cost
0.02268
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
0.00%
Matches word count
n/a
neededClean
false
words
532
33.97%
Dialogue to Total Word Ratio
Ratio: 54.65%, Deviation: 4.65%
neededClean
false
wordsTotal
538
wordsDialogue
294
16.9870%