NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-09-01KKAC37JTBCAAKZYY3D7SCA7E
Run Details
Model:
GPT-5.4 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
20.2s
Cost:
$0.017790
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
6.09%
Matches word count
(details)
95.55%
Dialogue to Total Word Ratio
(details)
50.8207%
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
“Molly, are you certain this is ethical?” “Bill, if they wanted to keep the presents, they shouldn’t have stacked them beneath a banner reading MAGIC HAPPENS HERE.” “Children happen there.” “Temporary people, Bill. Sticky, loud, temporary people.” “Still, robbing a birthday feels low.” “Lower than stealing from magicians? Because last week you stole a rabbit.” “It was a loan.” “The rabbit signed nothing. Listen: clown arrives at two, cake at three, sugar crash at four. We enter as exhausted parents.” “I can’t look exhausted on command.” Molly sighed. “You always look mildly disappointed. That’s suburban enough.” “What about the dog?” “Bribed with sausage.” “The pony?” “Intimidated by confidence.” “The birthday boy?” “Distracted by bubbles.” “The mothers?” “Bill, nobody notices anything at a children’s party. It’s just shrieking and beige hummus.” Bill adjusted his tie. “And if we’re caught?” “We say, ‘We’re the entertainment.’” “We don’t have an act.” “We have pockets.” “That’s not an act.” “It is if you do it with jazz hands.” A balloon popped outside. Bill flinched. “That sounded like destiny misfiring.” “No, that’s a balloon. Destiny makes a wetter noise. Come on, partner. We take the gifts, leave the card table, and vanish before the piñata gets political.” “Fine. But if I end up wearing a balloon hat, I’m retiring.” “Bill, if you survive the glitter, you’ll be legendary.”
created_at
"2026-03-09T22:39:44.746Z"
3/9/2026, 10:39:44 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
17338
moderation_latency
(null)
generation_time
20046
tokens_prompt
77
tokens_completion
458
native_tokens_prompt
66
native_tokens_completion
1175
native_tokens_completion_images
0
native_tokens_reasoning
840
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
0.01779
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_0bca8ce0b6721b9c0169af4c30d28c8195b7c96cc22d754174"
is_byok
false
latency
236
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-1773095984-zFy3SDafTZuoANpbvHig"
upstream_id
"resp_0bca8ce0b6721b9c0169af4c30d28c8195b7c96cc22d754174"
total_cost
0.01779
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
6.09%
Matches word count
n/a
neededClean
false
words
223
95.55%
Dialogue to Total Word Ratio
Ratio: 94.62%, Deviation: 4.62%
neededClean
false
wordsTotal
223
wordsDialogue
211
50.8207%