NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-18-01KM08M34F2CVQG2W9WWW5D7ST
Run Details
Model:
GPT-5.4 Nano (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 200 words with 10% dialogue
Duration:
5.0s
Cost:
$0.000954
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
43.38%
Matches word count
(details)
100.00%
Dialogue to Total Word Ratio
(details)
71.6887%
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
Bill crouched beside the minivan, adjusting a suction-cup mustache he’d stolen from a Halloween store. Molly checked the map, then checked the snack table nearby, as if the cookies might bribe the security system. The house belonged to a kid named Preston, whose parents had advertised “all ages” on the invitation, which Bill interpreted as “all exits, please.” He tapped his earbud; only the faint jingle of party favors answered back. “Molly, quiet. The clown’s nose has lasers,” Bill said. Molly rolled her eyes, pulled a glittery lockpick from her sock, and demonstrated its aerodynamic qualities by flicking it into the air. It landed perfectly in a potted fern, somehow looking intentional. She leaned toward the front door, listening to the muffled squeals of toddlers sprinting in circles. “Relax, Bill. Tonight we win, and kids love theft,” Molly whispered. Bill grinned, then remembered the part where they had to look like harmless helpers. He practiced his “friendly uncle” smile until it resembled a warning label. Molly arranged their gear into a decoy gift bag labeled THANKS, which immediately attracted a nosy balloon animal. Bill whispered to the balloon to stand guard. The balloon nodded, because balloons never argue. Together they drifted toward the doorstep, stepping carefully over chalk arrows that pointed toward the presents like a target.
created_at
"2026-03-18T10:42:20.447Z"
3/18/2026, 10:42:20 AM
model
"openai/gpt-5.4-nano-20260317"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3031
moderation_latency
(null)
generation_time
4742
tokens_prompt
77
tokens_completion
355
native_tokens_prompt
66
native_tokens_completion
753
native_tokens_completion_images
0
native_tokens_reasoning
462
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
0.00095445
router
(null)
provider_responses
0
endpoint_id
"0c835f2e-c18d-4e8c-b245-e1e3bd08b97f"
id
"resp_0e94b31e67351cf60169ba818c85a08190b818d271ca509a91"
is_byok
false
latency
151
model_permaslug
"openai/gpt-5.4-nano-20260317"
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-1773830540-goLDQpdfNhwHj6iGk9hp"
upstream_id
"resp_0e94b31e67351cf60169ba818c85a08190b818d271ca509a91"
total_cost
0.00095445
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
100.00%
Dialogue to Total Word Ratio
Ratio: 10.55%, Deviation: 0.55%
neededClean
false
wordsTotal
218
wordsDialogue
23
71.6887%