NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-06-01KK1HCQC44Q260CNCC5BCTKWS
Run Details
Model:
GPT-5.4 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
11.2s
Cost:
$0.005130
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
43.38%
Matches word count
(details)
36.79%
Dialogue to Total Word Ratio
(details)
40.0831%
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, tell me again why we’re robbing a children’s party.” “Because, Molly, tiny people get spectacular loot. Remote-control dinosaurs, glitter scooters, suspiciously expensive educational tablets.” “Educational tablets? We’re stealing arithmetic now?” “We’re stealing opportunity. Also batteries. Rich parents love batteries.” “This is low, Bill.” “Lower than robbing a bank?” “Morally, yes. Architecturally, no.” “Listen, it’s perfect. Noise, chaos, cake fog. Nobody notices two extra adults if one carries balloons.” “I hate balloons. They squeak judgment.” “Then wear the clown glasses.” “I’d rather confess to every crime I haven’t committed.” “Fine. We go in as the magician’s assistants.” “There’s a magician?” “There will be if I finish this cape.” “You’re sewing felt stars onto a bathrobe.” “Children are idiots.” “They are terrifyingly observant, actually. One six-year-old once spotted my fake pearl earring from across a bouncy castle.” “Then we distract them.” “With what?” “A treasure hunt.” “At a party already containing treasure.” “Exactly. We hide the presents in our van for ‘safekeeping.’” “You hear how that sounds?” “Professional.” “It sounds like kidnapping for dolls.” “No dolls. Electronics first, then premium Lego.” “And if parents ask questions?” “I’ll say, ‘Madam, your son won a mystery prize.’” “What mystery prize?” “A valuable lesson about impermanence.” “Bill.” “What?” “I hate that this might work.” “You hate brilliance in all its forms.”
created_at
"2026-03-06T12:19:08.812Z"
3/6/2026, 12:19:08 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
508
moderation_latency
(null)
generation_time
11177
tokens_prompt
77
tokens_completion
489
native_tokens_prompt
66
native_tokens_completion
331
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
0.00513
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_00a075d82ff7714b0169aac63cdf2c8194b39a0fd04139b9b4"
is_byok
false
latency
161
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-1772799548-gYNVNQgi50aNH33jnCtw"
upstream_id
"resp_00a075d82ff7714b0169aac63cdf2c8194b39a0fd04139b9b4"
total_cost
0.00513
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
43.38%
Matches word count
n/a
neededClean
false
words
217
36.79%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 10.00%
neededClean
false
wordsTotal
220
wordsDialogue
220
40.0831%