NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-18-01KM09653N35HJVPY0W8Z81M9J
Run Details
Model:
GPT-5.4 Nano
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
2.1s
Cost:
$0.000468
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
9.61%
Matches word count
(details)
36.79%
Dialogue to Total Word Ratio
(details)
23.1980%
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
“Molly, you’re sure the front door is unlocked?” “Bill, I’m sure. Uncle Randy always forgets his own keys when he drinks punch.” “Punch. Great. So we break in, grab the presents, and leave before the piñata police show up.” “Mmm, piñata police. That’s not a thing. Also, we’re not touching the piñata.” “Why not? It’s basically a piñata-shaped snack.” “Because last time you ‘casually’ hugged the mascot and got glitter in your eyebrows for a week.” “Glitter is decorative. Like stolen goods.” “Shh. Hear that?” “That’s the sound of parental vigilance. It’s louder than my conscience.” “Your conscience is missing. We need the presents, not your imaginary morals.” “Okay, plan: we sneak in, we grab everything under the tree, and we—” “And we replace the tags with blank ones so no kid notices.” “You want to counterfeit Christmas receipts?” “Not counterfeit. Just… educational.” “Educational theft. That’s worse than I thought.” “Quiet. I’m seeing movement.” “Is it the host? Or a toddler with a weaponized cupcake?” “Both are possible. Stay behind me.” “Do I get to steal the big one?” “The big one is probably a Lego set.” “Perfect. I love bricks. They’re like stealing, but with dignity.” “Dignity is optional. Kids are not.” “Fine. We steal carefully.” “Bill, you just called it ‘carefully’ out loud.” “Aw, come on—children’s parties have ears.”
created_at
"2026-03-18T10:52:12.291Z"
3/18/2026, 10:52:12 AM
model
"openai/gpt-5.4-nano-20260317"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
461
moderation_latency
(null)
generation_time
1734
tokens_prompt
77
tokens_completion
422
native_tokens_prompt
66
native_tokens_completion
364
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
0.0004682
router
(null)
provider_responses
0
endpoint_id
"0c835f2e-c18d-4e8c-b245-e1e3bd08b97f"
id
"resp_098570101c9ef4b00169ba83dc5ec88194a167814f8a85dacd"
is_byok
false
latency
200
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-1773831132-EYFeiLUbLGrbBRH8iXGp"
upstream_id
"resp_098570101c9ef4b00169ba83dc5ec88194a167814f8a85dacd"
total_cost
0.0004682
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
9.61%
Matches word count
n/a
neededClean
false
words
222
36.79%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 10.00%
neededClean
false
wordsTotal
223
wordsDialogue
223
23.1980%