NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX41612M9F0PABTJ22WQPYB9
Run Details
Model:
GPT-5.6 Terra (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
15.5s
Cost:
$0.017430
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.04%
Dialogue to Total Word Ratio
(details)
0.0185%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 a crayon-colored invitation across the café table. A unicorn wearing sunglasses beamed up at them. “Three o’clock Saturday,” she said. “Poppy Winkle’s seventh birthday. Bouncy castle, magician, cake, and, according to the mother’s public announcement on social media, ‘mountains of presents.’” Bill adjusted his wool cap. “Public announcement? That’s practically an engraved invitation.” “It is an invitation.” “To the party?” “To the crime, Bill. Keep up.” He peered at the card. “Dress code: glittery. We don’t own glittery.” “I own a silver jacket.” “You wore that to rob the bowling alley.” “And did anyone suspect me?” “Yes. The manager said, ‘Why is that woman dressed like a disco spaceship carrying fourteen pairs of rented shoes?’” “But he couldn’t prove anything.” “He had a photograph.” “Blurry photograph.” “Your jacket reflected the flash.” Molly tapped the invitation. “Focus. We enter as entertainment.” Bill brightened. “I can juggle.” “You can drop three oranges in a rhythm.” “That is essentially juggling.” “It is essentially breakfast failing.” “What about balloon animals?” “You made a balloon dachshund once.” “It had a tail.” “It had one leg and the face of a disappointed eel.” Bill considered this. “Children love eels.” “Children love presents. That’s why we’re taking them.” A waitress set down two coffees and stared at the unicorn invitation. “Planning something nice?” she asked. “Yes,” Molly said smoothly. “We’re volunteering at a children’s celebration.” Bill smiled too widely. “We adore joy.” The waitress left quickly. “See?” Molly said. “Natural.” “You said ‘celebration’ like a funeral director.” “I panicked.” “You’re the planner. You don’t panic.” “I panicked elegantly.” Bill leaned closer. “What’s the escape plan?” “We don’t call it an escape plan.” “What do we call it?” “A departure with gifts.” “Very classy. And how do we depart?” “In the van.” “The van with ‘Bill’s Carpet Cleaning’ painted on the side?” “That van is invisible.” “It is turquoise.” “Exactly. Nobody notices turquoise.” “Everybody notices turquoise. It’s the color of dental offices and regret.” Molly took a sip of coffee. “Then we cover the sign.” “With what?” “A banner.” “What kind of banner?” “Something festive.” Bill’s eyes widened. “We could say ‘Princess Pony Party Services.’” “We have no ponies.” “We have you.” “I am not a pony.” “You’re bossy, you bite when annoyed, and your hair is approximately chestnut.” “My hair is auburn.” “Princess Auburn Pony. Very premium.” Molly pinched the bridge of her nose. “We are not involving live animals, fake animals, or you in a mane.” “Fine. I’ll be a magician.” “You don’t know magic.” “I know one trick.” “The coin behind the ear?” “Yes.” “The one where you accidentally put the coin in your own ear?” “It was a rehearsal.” “For a doctor.” Bill lowered his voice. “What if the children recognize us?” “From where?” “The bowling alley.” “Bill, children do not maintain a criminal registry.” “They remember faces. Tiny, judgmental faces.” “They’ll be distracted by cake.” “Cake makes people careless,” Bill agreed. “That’s science.” Molly folded the invitation. “We blend in, compliment the parents, admire the decorations, and collect the presents when everyone watches the magician.” “But if I’m the magician—” “You are not the magician.” “Could I be the clown?” “No.” “The pirate?” “No.” “The mysterious uncle?” “You already look like the mysterious uncle.” Bill looked offended. “I look trustworthy.” “You look like a man who sells fireworks from a canoe.” “That is a very specific insult.” “It’s because you have the face for it.” He nodded slowly. “All right. We go as caterers.” “Perfect.” “Do caterers have to cook?” “Absolutely not. We carry a tray, look harassed, and say ‘allergy-friendly’ whenever challenged.” Bill grinned. “Molly, that is brilliant.” “I know.” “And after we leave with the presents?” “We donate half anonymously.” Bill blinked. “Half?” “They’re children’s presents.” “Right,” he said. “Half.” Molly gave him a look. “Fine. All of them.” “Good.” “But I’m keeping one thing.” “What?” “The unicorn invitation.” “Why?” Bill tucked it into his coat. “It’s evidence of a perfect crime.” “It has your coffee ring on it.” “Then it’s evidence of breakfast failing.”
created_at
"2026-07-09T18:10:33.692Z"
7/9/2026, 6:10:33 PM
model
"openai/gpt-5.6-terra-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2985
moderation_latency
(null)
generation_time
15341
tokens_prompt
77
tokens_completion
1572
native_tokens_prompt
66
native_tokens_completion
1151
native_tokens_completion_images
0
native_tokens_reasoning
166
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
0.01743
router
(null)
provider_responses
0
endpoint_id
"3cc89000-ae34-4dde-9c7d-5ce009c8af0b"
id
"resp_0616ea34e29010a4016a4fe419bfd88197be9331421708a362"
is_byok
false
latency
233
model_permaslug
"openai/gpt-5.6-terra-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-1783620633-SPXA308Hd62cFQaSECyc"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783620633-4f5sovHige4BgMo2Gwdq"
upstream_id
"resp_0616ea34e29010a4016a4fe419bfd88197be9331421708a362"
total_cost
0.01743
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
676
0.04%
Dialogue to Total Word Ratio
Ratio: 82.57%, Deviation: 12.57%
neededClean
false
wordsTotal
677
wordsDialogue
559
0.0185%