NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-06-01KK1J5JJ2EYRA98QHBR5WDQKY
Run Details
Model:
GPT-5.4
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
11.1s
Cost:
$0.005145
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
36.79%
Dialogue to Total Word Ratio
(details)
18.3943%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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
“Bill, this is either genius or a prison sentence with balloons.” “Molly, every great plan sounds bad when you say it near a bakery.” “We are robbing a child’s birthday party.” “We are liberating unappreciated merchandise from sticky-fingered minors.” “They’re literally the intended recipients.” “And we are unexpectedly efficient redistribution specialists.” “How do you propose we get in?” “Simple. I’m the magician, you’re the face painter.” “I can’t paint faces.” “You can draw a star and look confident. Children respect confidence and glitter.” “I hate glitter. It’s dandruff with ambition.” “Focus. We arrive during cake. Parents are distracted photographing sugar-drunk offspring.” “And then?” “We load the presents into the rabbit crate.” “You own a rabbit crate?” “I own several. Don’t derail the briefing.” “What if a child asks me for a unicorn?” “Paint a horse and say inflation hit the horn market.” “That’s monstrous.” “So is your tone. Then we slip out through the garden gate.” “With twenty presents?” “Eighteen. Two are always socks.” “You’ve studied this.” “I respect my craft.” “What about the clown?” “Bribed.” “With what?” “Cigarettes and emotional validation.” “Bill, this feels karmically radioactive.” “Molly, last week we stole from a duke.” “Yes, but he was forty-seven and deserved his curtains taken.” “So tonight we rob the tiny rich.” “If a six-year-old cries, I’m leaving.” “Fine. We only take the expensive-looking ones.” “That is somehow worse.” “That is called ethics.”
created_at
"2026-03-06T12:32:43.084Z"
3/6/2026, 12:32:43 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
541
moderation_latency
(null)
generation_time
11036
tokens_prompt
77
tokens_completion
499
native_tokens_prompt
66
native_tokens_completion
332
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)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"completed"
usage
0.005145
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_06854b0650dcc0000169aac96b253c819489757a1b84b34a67"
is_byok
false
latency
225
model_permaslug
"openai/gpt-5.4-20260305"
provider_name
"OpenAI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.0; linux; x64))"
http_referer
(null)
api_type
"completions"
id
"gen-1772800363-cZdQnvAiHlg5i1rZPfjz"
upstream_id
"resp_06854b0650dcc0000169aac96b253c819489757a1b84b34a67"
total_cost
0.005145
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
233
36.79%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 10.00%
neededClean
false
wordsTotal
239
wordsDialogue
239
18.3943%