NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHKAMK1EATYY1T30JRPW4S21
Run Details
Model:
GPT-5 Mini
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
1m 12s
Cost:
$0.013870
Provider:
OpenAI
Model input
User/Human
You are a author of comedic heist books. Please write a 500 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
They sat in the cramped van like conspirators sharing a picnic of sticky notes and stale coffee. Molly outlined the timeline on a greasy napkin, drawing arrows between the clown's entrance and the cake ceremony. Bill traced escape routes with a finger and tasted the snack table he planned to credit as alibi. Outside, little shoes thudded and a radio played a pop song that would become their cover track. The party smelled of frosting and rubber and the efficient anonymity of disposable plates. Molly pinched her lip and rehearsed small talk lines: "Oh, we were just cleaning up" and "Sorry, wrong table," until the phrases sounded like practiced keys. Bill adjusted a pair of cheap latex gloves, tested a zipper, and smiled at the absurdity of being both courteous and criminal. They were not monsters, they agreed; they were managers of entropy, relocating goods from temporary joy to their particular pockets. A paper crown lay on the dashboard, a poor monarch abandoned after a coronation, and Molly considered leaving it on the birthday chair as proof they had not been cruel. Light shifted as the sun moved along the rooftops and she tapped the napkin: fifteen minutes from clown to van, three minutes in the kitchen, two to swap bags. Their humor was precise, a steadying ritual, and when the bubble machine plan made them laugh, they both let out small, conspiratorial sighs. Tonight felt like a small, polite war fought over glitter and squeakers. They were oddly sentimental. Molly: We go in during the clown act, while everyone watches the balloon animals. Bill: And the cake gives us cover with smoke and movement. Molly: Kids scream, parents sing, we glide like shadows between paper hats. Bill: Start in the coatroom, empty hands, pretend to be cleanup crew later. Molly: Bags that look like party supplies, full of streamers but really heavy. Bill: No toys left behind. Teddy bears, game boxes, bikes — take the lot. Molly: We must be quick, quiet, polite if stopped. Smile, apologize, leave. Bill: Polite burglars, like hospital volunteers who move faster. Molly: Hide behind the piñata stand, scoop the pile, shuttle through the pantry. Bill: Use pockets, foldables, dismantle big boxes into flat envelopes. Molly: Label everything "leftovers" and shove in the bins of the caterer. Bill: The caterer owes me. Trade for dessert credits. Molly: What about the birthday kid? We leave one present, a decoy with a note. Bill: "From Santa with love, returned by mistake." Subtle. Molly: Cameras? Bill: Two, in corners, blind spots mapped. We dance in between. Molly: Noise plan? Bill: Trigger the bubble machine. Kids will chase bubbles, creating chaos. Molly: Bubble machine and the clown's confetti cannon. Perfect. Bill: Exit through the kitchen, through the service door, and into the van. Molly: Van with fake caterer logo. We swap logos, leave no tracks. Bill: We split by neighborhood, fences, safe houses. No bragging, no calls. Molly: One rule: no targeting photos on mantels—keep people's memories. Bill: Disappear.
created_at
"2026-02-16T13:37:15.174638+00:00"
2/16/2026, 1:37:15 PM
model
"openai/gpt-5-mini-2025-08-07"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
45683
moderation_latency
(null)
generation_time
71928
tokens_prompt
77
tokens_completion
1340
native_tokens_prompt
66
native_tokens_completion
6927
native_tokens_completion_images
0
native_tokens_reasoning
6208
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
0.0138705
router
(null)
provider_responses
0
id
"resp_0affe3e87603c4d80169931d42c02c8193accff708884ffe13"
status
200
is_byok
false
latency
173
endpoint_id
"c4f66d01-20b0-4c27-a225-438ea22fda43"
provider_name
"OpenAI"
model_permaslug
"openai/gpt-5-mini-2025-08-07"
api_type
"completions"
id
"gen-1771248962-s9FQcIeAkoZwdYBTk0ut"
upstream_id
"resp_0affe3e87603c4d80169931d42c02c8193accff708884ffe13"
total_cost
0.0138705
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
99.9990%
Matches word count
n/a
neededClean
false
words
501
0.0000%
Dialogue to Total Word Ratio
Ratio: 3.39%, Deviation: 46.61%
neededClean
false
wordsTotal
501
wordsDialogue
17
49.9995%