NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KBRVZ1MYRKHQ5XB6G486W4
Run Details
Model:
DeepSeek V4 Pro 0813 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
17.5s
Cost:
$0.003942
Provider:
DigitalOcean
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
Overall
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 storage unit smelled of stale bubble mixture and a decade of abandoned birthday candles, which was fitting because Molly was hunched over a hand-drawn map of a suburban backyard, her finger tracing the route from the hedge to the inflatable bouncy castle like a general planning a beach landing. Bill sat on an overturned plastic crate, eating a bag of cheese puffs with the focused intensity of a man who had never been told to stop. Molly tapped the map. The X marked the gift table, a sturdy folding arrangement draped in a paper tablecloth printed with smiling dinosaurs. She had counted at least twelve presents during her reconnaissance, which meant twelve chances for something expensive and resellable. Bill licked orange dust from his thumb. "So we just walk in wearing party hats and grab the presents?" "No. We don't just walk in wearing party hats. That's insane. We walk in wearing party hats and we're holding a piñata." "What's the piñata for?" "It's a cover. Nobody questions two adults carrying a piñata at a children's party. We look like we're either the entertainment or extremely lost relatives." Bill considered this. "Do we have to fill the piñata with candy? Because I bought a bag of candy for morale." Molly closed her eyes for a brief second. "Bill. The piñata is empty. That's the whole point. It's a hollow papier-mâché horse that we will hold above our heads while we cross the yard, and if anyone asks, we're taking it to the garage to fix the string." "But why is the string broken?" "It isn't. The string is perfect. The story is the string is broken. That's the lie, Bill. The lie is the string." Bill nodded slowly, then stopped. "What about the clown?" Molly froze. "What clown?" "I saw a flyer at the mailbox. It says 'Bouncy Castle, Face Painting, and Chuckles the Clown.' Chuckles has a red nose and a hat with a flower that squirts water. I don't trust him." Molly leaned back. "Nobody trusts Chuckles. That's the first rule of birthday heists. Chuckles is a wild card. But he's also a professional distraction. While every child is screaming with joy or terror at Chuckles, we slide in, grab the presents, and slide out." "Sliding in a bouncy castle area is risky. There could be discarded juice boxes." Molly pointed at him with a cheese puff-stained finger. "That's why we wear sneakers with good grip. Not those ridiculous slip-on loafers you like." Bill looked down at his loafers. "They're comfortable. And they have tassels." "The tassels are a liability in wet grass. You'll go down like a giraffe on roller skates." "That's a vivid image." "I've been preparing for years, Bill. Now focus. We arrive at two-fifteen, when the sugar crash begins and parents start checking their phones for early pickup excuses. We park two streets over and approach through the gap in the fence behind the compost bin." "I hate the compost bin. It smells like a swamp's armpit." "That's the smell of anonymity. We put on the hats, we grab the piñata, we stride through the yard like we own the bouncy castle itself. If anyone waves, you wave back with the hand holding the piñata. Make it look jaunty." "Jaunty. Got it." "At the gift table, you drop the empty piñata and fill the piñata with presents instead. It has a flap on the bottom. I sewed it on." "With what thread?" "Embroidery floss. It'll hold twelve presents." "What if one of the kids sees us and wants the piñata?" Molly smiled a thin, professional smile. "Then we tell them it's full of bees. Kids respect bees." Bill thought about that. He nodded, impressed. Then he reached into the cheese puff bag and frowned. "I ate all the morale candy." "Of course you did." Molly stood and cracked her knuckles. "Get the party hats. The good ones with the elastic strings. We're going to be the best-dressed present thieves in the county."
created_at
"2026-09-03T10:07:00.607Z"
9/3/2026, 10:07:00 AM
model
"deepseek/deepseek-v4-pro-20260813"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
963
moderation_latency
(null)
generation_time
17202
tokens_prompt
75
tokens_completion
1053
native_tokens_prompt
67
native_tokens_completion
973
native_tokens_completion_images
(null)
native_tokens_reasoning
65
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
service_tier
(null)
usage
0.00394152
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788430020-D8RY1UXBMzQjFXaCsXHb"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788430020-EA46luIVuP624ck5YPPa"
upstream_id
"chatcmpl-3366b1aa-5137-47ff-9409-049907e32449"
provider_responses
0
endpoint_id
"928ab2e9-a85f-427f-adcb-15b2394c70d5"
id
"chatcmpl-3366b1aa-5137-47ff-9409-049907e32449"
is_byok
false
latency
963
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"DigitalOcean"
status
200
total_cost
0.00394152
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DigitalOcean"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
672
0.00%
Dialogue to Total Word Ratio
Ratio: 71.09%, Deviation: 41.09%
neededClean
false
wordsTotal
678
wordsDialogue
482
0.0000%