NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37HD3X1ESNPXZ94N75YA8PR
Run Details
Model:
GPT-6 Sol (Reasoning, Medium)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
22.8s
Cost:
$0.007762
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
97.63%
Matches word count
(details)
10.10%
Dialogue to Total Word Ratio
(details)
53.8631%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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 floor plan across the café table, pinning its corners with two sugar bowls and a stolen traffic cone. Across the street, balloons bobbed above the entrance to the Happy Hippo Play Barn. Every child who went in carried a present. Every parent who came out looked relieved to have handed one over. “Thirty-seven presents,” Molly said. “One door, one loading bay, and a mascot who takes a break at three.” Bill studied the drawing. “Why have you marked the ball pit with a skull?” “Because you still owe me an apology for Tuesday.” “Tuesday was a misunderstanding. I thought you said hide the diamonds in the ball pit.” “I said don’t.” Bill adjusted his fake moustache. It was made of felt and sat crookedly above his real one. He had dressed as a magician, reasoning that nobody questioned a man in a cape carrying a large empty trunk. Molly had dressed as his assistant, reasoning that somebody needed to question him. At three fifteen, according to Molly’s reconnaissance, the guests would gather around the cake. She would trigger the sprinkler in the puppet theatre, causing a distraction. Bill would wheel the presents through the loading bay. By three twenty, they would be rich, or at least surrounded by expensive toys they could sell. “Isn’t the sprinkler connected to the fire alarm?” Bill asked. “Only if there’s a fire.” “Then what are you triggering?” “A convincing puppet emergency. You’ll see.” Bill folded his arms. “I would prefer to see it before we’re inside.” They crossed the road with the trunk rattling behind them. A woman at the play barn door handed Bill a paper crown and Molly a clipboard. Before either could object, she hurried them into a room full of bunting, shrieking children, and a cake shaped like a suspiciously accurate hippopotamus. “There you are!” she cried. “The entertainment agency said you’d be early.” Molly looked at the clipboard. It contained a schedule, a list of food allergies, and one instruction underlined three times: DO NOT LET THE MAGICIAN NEAR THE CAKE. Bill saw it too. “That seems unnecessarily specific.” The woman vanished into the kitchen. Twenty-seven children turned toward Bill. He had once escaped a museum through a ventilation shaft the width of a bread bin. This was worse. He opened the trunk, hoping inspiration might be inside. It was not. “Do a trick,” Molly whispered. “With what?” “The moustache. Make it disappear.” Bill peeled it off and tucked it up his sleeve. The children gasped. Molly, surprised, applauded. Bill bowed and knocked his paper crown into the frosting. For forty glorious minutes, they performed increasingly terrible magic. Molly produced a balloon from behind a chair. Bill guessed the colour of a crayon he was holding in plain sight. By the end, the children were cheering and the presents were untouched. Outside, Molly counted the notes the woman had pressed into her hand. “Eighty pounds,” she said. “And no prison,” Bill replied. Molly considered the traffic cone under her arm. “We’re keeping this.”
created_at
"2026-09-23T16:25:57.417Z"
9/23/2026, 4:25:57 PM
model
"openai/gpt-6-sol-20260922"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4615
moderation_latency
(null)
generation_time
22717
tokens_prompt
75
tokens_completion
944
native_tokens_prompt
66
native_tokens_completion
763
native_tokens_completion_images
0
native_tokens_reasoning
113
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
0.007762
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790180757-DQrtjifqWPesDgVR42gn"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790180757-AnMNkHgDTwu27uGKbcbW"
upstream_id
"resp_03d8c9de2ee3847e016ab3fd95878087d1800f6ab129d032fe"
provider_responses
0
endpoint_id
"1cfc7d9d-4404-4b8e-9ee4-58ae45c9dcd4"
id
"resp_03d8c9de2ee3847e016ab3fd95878087d1800f6ab129d032fe"
is_byok
false
latency
667
model_permaslug
"openai/gpt-6-sol-20260922"
provider_name
"OpenAI"
status
200
total_cost
0.007762
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
97.63%
Matches word count
n/a
neededClean
false
words
507
10.10%
Dialogue to Total Word Ratio
Ratio: 23.23%, Deviation: 6.77%
neededClean
false
wordsTotal
508
wordsDialogue
118
53.8631%