NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-09-01KKAC79Q6QWB8RRMHRQV5C3ZC
Run Details
Model:
GPT-5.4 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
11.0s
Cost:
$0.006000
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, this is either genius or the sort of idea that gets us pelted with cupcakes.” “Cupcakes wash off. Jewels don’t usually come wrapped in dinosaur paper, Molly.” “They do tonight. Twelve gifts, one bounce castle, three exhausted parents, and a magician with the eyes of a tax inspector.” “I’ve accounted for the magician.” “How?” “I’m his rabbit.” “You are not.” “I borrowed the costume. Ears, tail, existential shame, the lot.” “Fine. I distract the adults?” “With what?” “A clipboard and purpose. Nobody questions a woman with purpose.” “And I sweep the presents into the catering van.” “The van says ‘Princess Sparkle Entertainment.’” “Exactly. Who suspects a glitter unicorn?” “Bill, last time you said that, we escaped in a petting-zoo trailer.” “And flawlessly.” “We smelled like alpaca for weeks.” “Focus. At cake-cutting, the children surge forward, the parents produce cameras, and civilization collapses.” “I hate how right you are.” “You’ll love this part. We leave a decoy pile.” “Of what?” “Educational books.” “That’s not a decoy. That’s a declaration of war.” “Children never unwrap those first.” “One child will. There’ll be screaming.” “There’s already screaming. It’s a children’s party.” “What about the clown?” “Bribed.” “With cash?” “With a better balloon pump.” “Bill, you magnificent idiot, let’s rob some minors.” “Metaphorically, Molly. We’re stealing from their rich aunties.”
created_at
"2026-03-09T22:41:57.999Z"
3/9/2026, 10:41:57 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3479
moderation_latency
(null)
generation_time
10918
tokens_prompt
77
tokens_completion
573
native_tokens_prompt
66
native_tokens_completion
389
native_tokens_completion_images
0
native_tokens_reasoning
59
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
0.006
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_0e4eca0d6807abdf0169af4cb6121081958c7f9b8ca679768a"
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-1773096117-aFowxYv17QrOKMbTdIIT"
upstream_id
"resp_0e4eca0d6807abdf0169af4cb6121081958c7f9b8ca679768a"
total_cost
0.006
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
219
wordsDialogue
219
40.0831%