NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-06-01KK1J5NH8SJV23NNF5RNX9V11
Run Details
Model:
GPT-5.4 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 200 words with 10% dialogue
Duration:
9.7s
Cost:
$0.004515
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
35.00%
Matches word count
(details)
98.16%
Dialogue to Total Word Ratio
(details)
66.5790%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 10% 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 crouched behind the privet hedge, black turtleneck collecting confetti blown from Number Twelve’s front garden. Beyond the gate, balloons bobbed like cheerful sentries, and the shriek of six-year-olds rose and fell with the alarming rhythm of a police siren. Bill, beside her, unfolded a crayon-sketched map he’d lifted from the birthday girl’s little brother and studied it with the solemnity of an admiral invading a nation made of frosting. “Front door’s chaos,” Molly whispered. “We go through the bouncy castle.” Bill adjusted his wool cap and peered over the hedge. “Bold. Undignified. I like it.” A clown on stilts tottered past the window carrying a tray of blue jelly. In the back garden, a pirate magician was losing an argument with a rabbit. Near the hall table, glitter-wrapped presents sat in a glorious mound, practically humming to Molly like a choir of unattended banknotes. She checked her watch. Three minutes until cake, the sacred lull when every child in Britain became temporarily stationary and sugared into reverence. Bill produced two party hats from his coat as if revealing forged passports. Molly took the hat with a grim nod. The elastic snapped under her chin with criminal authority. Then they slipped from the hedge, smiling like exhausted cousins, and advanced on the gate through a minefield of bubbles.
created_at
"2026-03-06T12:32:46.128Z"
3/6/2026, 12:32:46 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
507
moderation_latency
(null)
generation_time
9615
tokens_prompt
77
tokens_completion
353
native_tokens_prompt
66
native_tokens_completion
290
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
0.004515
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_0accfd7a2c299a9d0169aac96e2e488193aa1ba2f1d8f35f26"
is_byok
false
latency
153
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-1772800366-LpJzvZur1gSWK9eAohF7"
upstream_id
"resp_0accfd7a2c299a9d0169aac96e2e488193aa1ba2f1d8f35f26"
total_cost
0.004515
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
35.00%
Matches word count
n/a
neededClean
false
words
218
98.16%
Dialogue to Total Word Ratio
Ratio: 6.31%, Deviation: 3.69%
neededClean
false
wordsTotal
222
wordsDialogue
14
66.5790%