NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M36P5801FNW4D6VFPK60HW4K
Run Details
Model:
GPT-6 Luna
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
12.1s
Cost:
$0.000472
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
4.40%
Dialogue to Total Word Ratio
(details)
2.1990%
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
“We go in at three,” Molly whispered. “Three in the morning?” Bill asked. “Three in the afternoon.” “That’s when the children are awake.” “Exactly. Nobody suspects a heist during a magic show.” Bill peered through the hedge at the community hall. A banner flapped above the door: HAPPY BIRTHDAY, OLIVER! Beneath it sat a mountain of brightly wrapped presents. “I count twenty-seven,” he said. “I count twenty-eight.” “That small one is a chair.” “It has a bow.” “Maybe Oliver is getting a chair.” “Then we steal the chair too.” Bill adjusted his black shirt, which had a cartoon dinosaur on it. “I thought we agreed on disguises.” “This is my disguise.” “You’re dressed like a children’s party guest.” “I’m dressed like a sophisticated children’s party guest.” “Your hat has a propeller.” “It’s for air circulation.” Molly handed him a paper plate. “Here. You’re my cousin.” “Do I have a name?” “Cousin Bill.” “That’s suspiciously close to my real name.” “It’s a party. Nobody asks questions at a party. They’re too busy eating cake and being emotionally manipulated by balloons.” Bill looked at the hall again. “What’s the plan?” “We enter through the front door, blend in, locate the presents, and remove them one at a time.” “One at a time? There are twenty-eight.” “We’ll use the cake trolley.” “Does the hall have a cake trolley?” “It will when I bring one in.” “You brought a trolley?” “I brought two.” “Why two?” “In case one gets lonely.” A burst of music came from inside. Children cheered. Molly smiled. “Now,” she said. They marched through the door. A woman with a clipboard blocked their path. “Names?” she asked. “Cousin Molly,” Molly said. “Cousin Bill,” Bill added. “From which side of the family?” “The left,” said Molly. The woman blinked. “The left?” “His father’s side,” Bill said quickly. “Very complicated. Lots of cousins.” “Are you on the list?” “Probably,” Molly said. “I have a face people remember.” “I don’t,” Bill said. “That’s why I’m useful.” The woman studied them, then pointed toward the party. “Oliver’s parents are by the cake.” “Wonderful,” Molly said. “We love parents.” “I’m not sure I do,” Bill murmured as they passed. “Focus,” Molly whispered. “The gifts are beside the stage.” “Where’s the trolley?” “Behind the magician’s curtain.” “You brought a trolley into a party?” “I told the magician I was his assistant.” “You’re not wearing an assistant’s outfit.” “I’m wearing a hat with a spinning part. That’s basically a uniform.” A magician in a purple cape appeared beside them. “You’re late.” “Traffic,” Molly said. “You were supposed to help with the disappearing rabbit.” “Can the rabbit lift presents?” “No.” “Then we have different priorities.” Bill leaned toward the gift pile. “There’s a parent watching.” “Distract them.” “How?” “Ask about the cake.” Bill turned to a man holding a camera. “Is that cake gluten-free?” “I don’t know.” “Could you find out? My cousin is sensitive.” Molly slid a present onto the trolley. The magician gasped. “That’s not the rabbit!” “Not yet,” Molly said. Then Oliver appeared, wearing a paper crown. “Are you taking my presents?” Molly froze. Bill stopped the trolley. “Absolutely not,” Bill said. “We’re checking the wrapping,” Molly added. “For safety.” Oliver narrowed his eyes. “Why?” “Some wrapping paper is dangerous,” Bill said. “Very slippery.” Oliver considered this. “Can you help me open them?” Molly glanced at Bill. “Of course,” she said. “That was our plan all along.” “Then you can stay,” Oliver declared, “but you have to sing.” Bill looked horrified. “What song?” “Happy birthday.” Molly grabbed the trolley handle. “We’re doomed.” “Sing,” Oliver ordered. So they sang, badly and enthusiastically, while Oliver opened every present. And when the party ended, Molly and Bill left with nothing but two slices of cake, a propeller hat, and a deep understanding of why children were the most formidable security force in the world.
created_at
"2026-09-23T08:29:47.911Z"
9/23/2026, 8:29:47 AM
model
"openai/gpt-6-luna-20260922"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1972
moderation_latency
(null)
generation_time
12057
tokens_prompt
75
tokens_completion
1203
native_tokens_prompt
66
native_tokens_completion
930
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
0.0004716
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790152187-ulHV4Y1KPFSFtiQHy60j"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790152187-f86itRgaCRJppmUst12e"
upstream_id
"resp_0a5a83470fa05c0b016ab38dfc013887d180659bb1133e8a86"
provider_responses
0
endpoint_id
"05e94e02-b9c2-4bbb-ba55-4082ee9ad687"
id
"resp_0a5a83470fa05c0b016ab38dfc013887d180659bb1133e8a86"
is_byok
false
latency
937
model_permaslug
"openai/gpt-6-luna-20260922"
provider_name
"OpenAI"
status
200
total_cost
0.0004716
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
0.00%
Matches word count
n/a
neededClean
false
words
645
4.40%
Dialogue to Total Word Ratio
Ratio: 62.10%, Deviation: 7.90%
neededClean
false
wordsTotal
649
wordsDialogue
403
2.1990%