NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M36R3E4CHMD2RYGT36RM7MSZ
Run Details
Model:
GPT-6 Luna (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
2m 8s
Cost:
$0.006747
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
100.00%
Dialogue to Total Word Ratio
(details)
100.0000%
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 and Bill crouched behind a hedge, studying the rainbow balloon arch guarding the community hall like a very cheerful dragon with excellent dental coverage. “There are forty presents inside. I counted every ribbon.” “You counted ribbons?” “They were color coded. This is research, Bill, not some advanced craft class.” Bill unfolded a napkin map, which showed the hall, the kitchen, and a sketch of a duck wearing sunglasses, for reasons Molly refused to ask. “The duck is security?” “No, it’s the birthday boy. He drew this map.” “Then we should fear his artistic range, especially when he’s entirely unsupervised.” Their plan was simple: enter during musical chairs, locate the gift table, and exit before anyone noticed the suspiciously gift shaped sacks they had packed. “When the music stops, everyone sits. We grab gifts according to protocol.” “And if someone sits on a gift?” “We congratulate their excellent timing, politely.” Molly produced a black outfit, then reconsidered when the hall’s windows revealed her reflection looked exactly like a burglar in a children’s book, tonight, apparently. “Black is wonderfully stealthy.” “Black is alarming. I look like a cartoon burglar.” “A convincing one.” “That is not encouragement.” “I am trying here, honestly.” From inside came a burst of laughter, followed by a child’s trumpet solo that sounded like a goose arguing with a kettle during a thunderstorm. “Children are unpredictable. They detect adults pretending to enjoy games.” “I lost at pin the tail.” “Were you blindfolded?” “Emotionally, yes. I was simply overwhelmed.” A paper crown drifted beneath the door, landing at Bill’s feet; he picked it up and wore it, testing its authority with grave professional concern. “Your Majesty, the loot awaits.” “Then let our subjects party on.” “You are not king.” “Not with that attitude.” “The van has my spare crown.” The side entrance opened with a squeak, revealing a corridor full of streamers, toy dinosaurs, and one vigilant grandmother knitting furiously beside the punch bowl. “Good evening. We are entertainers.” “What do you entertain?” “Mostly ourselves, madam.” “Then you’ll be delighted by the party.” “We accept cash, snacks, or applause.” Before either could improvise further, a small boy marched up wearing a cape and asked whether their disguises came in smaller sizes for younger villains. “We’re the party magicians.” “Can you make presents disappear?” “We can make them reappear.” “With better wrapping?” “With honest receipts.” “Can you make dinosaurs dance?” While children debated dinosaur magic, Molly and Bill bundled presents into sacks; the grandmother recognized the famous balloon thieves, and Molly elbowed Bill quite firmly. “You’re the balloon thieves?” “Retired.” “We specialize in presents.” “He means helping with presents.” “Excellent. Carry these inside.” “And perhaps entertain the children during cake.” They carried all presents back to the hall’s gift table, where children surrounded them, demanding magic, balloon animals, and demonstration of the crown before cake. “We came to steal everything. But now we’re booked until cake.” “Our greatest heist is childcare.” “And the crown?” “Tax deductible, if we keep receipts.”
created_at
"2026-09-23T09:03:45.815Z"
9/23/2026, 9:03:45 AM
model
"openai/gpt-6-luna-20260922"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4680
moderation_latency
(null)
generation_time
127622
tokens_prompt
75
tokens_completion
1891
native_tokens_prompt
66
native_tokens_completion
13480
native_tokens_completion_images
0
native_tokens_reasoning
12797
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
0.0067466
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1790154225-z1aVQCorkEq8Df2m6Qbq"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790154225-0YTG6gpc2uTAtKfu6AuY"
upstream_id
"resp_0f4cc255dffc27ec016ab395f1e85887d1b4260a07d1ac86be"
provider_responses
0
endpoint_id
"05e94e02-b9c2-4bbb-ba55-4082ee9ad687"
id
"resp_0f4cc255dffc27ec016ab395f1e85887d1b4260a07d1ac86be"
is_byok
false
latency
898
model_permaslug
"openai/gpt-6-luna-20260922"
provider_name
"OpenAI"
status
200
total_cost
0.0067466
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
500
100.00%
Dialogue to Total Word Ratio
Ratio: 50.00%, Deviation: 0.00%
neededClean
false
wordsTotal
500
wordsDialogue
250
100.0000%