NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-09-01KKACDZG6XVT859PGQVT8WDX9
Run Details
Model:
GPT-5.4 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 200 words with 50% dialogue
Duration:
21.6s
Cost:
$0.022635
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
92.54%
Dialogue to Total Word Ratio
(details)
96.2705%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 50% 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 peered over the hedge at the riot of balloons, wobbling like nervous witnesses in the dusk. Bill adjusted his tie, which was printed with tiny ducks and somehow made him look more criminal. “Children’s parties,” Molly whispered. “Nature’s cruelest buffet.” “Focus,” Bill said. “We’re here for presents, not jelly.” “I can multitask. How many tiny aristocrats are inside?” “Eight, possibly nine if the birthday boy has a gloomy cousin.” “And the clown?” “Already defeated by face paint and his own despair.” Molly grinned. “Good. I hate professionals.” A shriek rose from the garden, followed by the unmistakable thud of a piñata meeting its destiny. Bill flinched. “We slip in, smile strangely, and leave richer,” he said. “With what cover?” “I’m the magician.” “You can’t do magic.” “I can make credibility disappear.” Molly straightened her cardigan. “Fine. I’m the aunt from Leeds.” “You’ve never been to Leeds.” “Exactly. I’ll be mysterious.” They crept toward the gate as a confetti cannon fired early, showering Bill in glitter. “Now I look festive,” he muttered. “No,” Molly said, patting his shoulder. “Now you look arrestable.” Molly smirked. “Cheer up. If we fail, there’s always cake, and prison parties are famously underdecorated too.”
created_at
"2026-03-09T22:45:36.912Z"
3/9/2026, 10:45:36 PM
model
"openai/gpt-5.4-20260305"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
20881
moderation_latency
(null)
generation_time
21166
tokens_prompt
77
tokens_completion
507
native_tokens_prompt
66
native_tokens_completion
1498
native_tokens_completion_images
0
native_tokens_reasoning
1193
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
0.022635
router
(null)
provider_responses
0
endpoint_id
"9ff5625c-403f-4d7f-b895-58ac7295062c"
id
"resp_0b5dc00ef38f1a740169af4d91000881909235f290885e767e"
is_byok
false
latency
220
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-1773096336-kXbdoDCow8pl63lj2VRE"
upstream_id
"resp_0b5dc00ef38f1a740169af4d91000881909235f290885e767e"
total_cost
0.022635
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
199
92.54%
Dialogue to Total Word Ratio
Ratio: 55.28%, Deviation: 5.28%
neededClean
false
wordsTotal
199
wordsDialogue
110
96.2705%