NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5DWESKT5Q749HX64QW9SM
Run Details
Model:
DeepSeek V3 (2025-03-24)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
27.9s
Cost:
$0.000698
Provider:
Novita
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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
The air in the dingy basement was thick with the scent of stale coffee and poorly concealed desperation. Molly tapped her fingers against the folding table, staring at the hand-drawn blueprint of the Wentworth estate like it held the secrets of the universe. Or at least the secret to getting a decent payday. Bill leaned back in his chair, legs propped up on the table, crunching on a stale bag of cheese puffs. "So, let me get this straight. We're robbing a kid's birthday party?" Molly scowled. "Not just any kid's party. *The* kid's party. Tommy Wentworth. His dad owns half the tech companies in the city, and that little brat gets presents worth more than our last three jobs combined." Bill wiped orange dust on his pants. "You sure this ain’t a step down from our usual gigs? I mean, last month we lifted a diamond necklace from a gala. Now we’re swiping action figures and gold-plated rattles?" "Think bigger, Bill. These aren’t just toys. We’re talking limited-edition *everything.* Signed baseballs, tiny designer suits, a solid gold pacifier—probably." She pointed at the blueprint. "And the best part? The security’s a joke. One rent-a-cop at the gate, a nanny who’s probably glued to her phone, and a dog that’s more fluff than fury." Bill squinted. "You sure about the dog?" "Positive. It’s a Pomeranian named Mr. Snuffles." He snorted. "Right, terrifying." Molly smirked. "Exactly. We slip in during the chaos—kids screaming, cake flying, parents desperately pretending they’re having fun. Grab the gifts, load ‘em up, and vanish before anyone notices." Bill scratched his chin. "What’s our exit strategy? We can’t exactly bolt through a crowd of sugar-crazed six-year-olds." "Ah, that’s the beauty of it. The Wentworths have a service entrance for caterers. No one pays attention to staff. We dress up like clown waiters, blend in, and boom—clean getaway." Bill groaned. "Clown waiters? You’re joking." "Foolproof disguise. No one looks a clown in the eye. It’s basically invisibility." Bill tossed the empty chip bag aside. "Alright, but if we get caught, I’m blaming you for the clown thing." Molly rolled her eyes. "Deal. Now, you handle the van. I’ll fake a catering order. And Bill?" "Yeah?" "Try not to eat all the evidence this time." He grinned. "No promises." The plan was stupid. The plan was reckless. The plan was *perfect.*
created_at
"2026-02-16T21:24:43.1999+00:00"
2/16/2026, 9:24:43 PM
model
"deepseek/deepseek-chat-v3-0324"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
777
moderation_latency
(null)
generation_time
27890
tokens_prompt
77
tokens_completion
644
native_tokens_prompt
69
native_tokens_completion
607
native_tokens_completion_images
(null)
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
"stop"
usage
0.00069847
router
(null)
provider_responses
0
id
"5b56d57c8d2436a737a000017989f12a"
status
200
is_byok
false
latency
777
endpoint_id
"3a816eb8-7a0b-4a9a-b497-e3a1efb349fc"
provider_name
"Novita"
model_permaslug
"deepseek/deepseek-chat-v3-0324"
api_type
"completions"
id
"gen-1771277054-0MObGcHc8rLqbYaWiwR7"
upstream_id
"5b56d57c8d2436a737a000017989f12a"
total_cost
0.00069847
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Novita"
Evaluation details
Result
Evaluator
Details
Meta Data
0.0000%
Matches word count
n/a
neededClean
false
words
402
0.0000%
Dialogue to Total Word Ratio
Ratio: 68.91%, Deviation: 38.91%
neededClean
false
wordsTotal
402
wordsDialogue
277
0.0000%