NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-10-01M25DDA5GT2ZYVZJM3BZWGAZH
Run Details
Model:
DeepSeek V4.1 Flash (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
45.4s
Cost:
$0.006793
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
86.38%
Matches word count
(details)
97.87%
Dialogue to Total Word Ratio
(details)
92.1246%
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
Molly spread the birthday invitation across the dashboard of a borrowed ice-cream van and clicked a unicorn pen. Across the street, the Wigglesworth house throbbed with balloons, screaming children, and a rented unicorn that was clearly a pony wearing a traffic cone. Bill sat beside her in a bear costume that smelled like wet carpet and regret. He held a clipboard covered in stickers because Molly believed every felony needed a theme. "The presents go to the playroom after cake," Molly said. "You push the piñata cart. I cut the lock." Bill scratched his bear ear. "And if a kid sees me?" "Kids see nothing. They're still learning object permanence." He flipped through the clipboard. "Forty children. That's not a party, it's a tiny mob with juice boxes." "Which is why we have maps, a distraction, and a bear costume with pockets." "The pockets are because you sewed them on while I slept." "Initiative." Across the street, a clown tripped over a croquet mallet and fell into the bounce house. The children screamed with joy. Molly checked the time. Eleven minutes until the magician's finale, six until cake, two until the nanny vanished behind the hydrangeas to vape. Bill practiced nodding like a ventriloquist with laryngitis, but the bear head made him look like a mascot who had just seen the hotel bill. Molly checked the van's rear doors, where forty empty duffel bags waited, each labeled with a child's name in glitter glue. Organization was important. The plan was simple, which worried Bill. Simple plans were usually the ones that ended with him zip-tied to a balloon arch. He reviewed the route again: through the catering tent, past the face-painting table, behind the magician's folding screen, then out the side gate before the clown re-inflated. Molly had marked the guards with tiny stickers: two retired clowns, a bounce-house attendant named Kyle, and a nanny who vaped like a chimney. The only wild card was a toddler with a water pistol. Molly insisted toddlers were not security, merely weather. "Weather with a water pistol," Bill muttered. Molly ignored him and adjusted her fake caterer name tag, which read Hello, My Name Is Also Molly. It was not her best alias. "It never is," Bill said. "Neither was Auntie Marge." "One more time," Molly said. "Catering tent. You left, me right. Piñata cart by the face-painting table. No cake." "What if the cake touches me?" "Then you're a bear with frosting. Keep moving." Bill sighed. "I hate this plan." "You hate every plan." "I hated the museum plan less." "The museum had lasers." "The museum had dignity." Molly folded the invitation and tucked it into her apron. "Dignity doesn't pay for a new transmission. Presents do. Now put on your happy bear face." Bill tried. The bear looked constipated. Molly opened the van door, and the sound of forty seven-year-olds singing about sharks rolled over them like a cheerful tsunami. They stepped into the sunshine, carrying empty bags and criminal ambition, ready to ruin a birthday in the most profitable way possible.
created_at
"2026-09-10T10:21:58.837Z"
9/10/2026, 10:21:58 AM
model
"deepseek/deepseek-v4.1-flash-20260910"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
671
moderation_latency
(null)
generation_time
45253
tokens_prompt
75
tokens_completion
10931
native_tokens_prompt
93
native_tokens_completion
11298
native_tokens_completion_images
(null)
native_tokens_reasoning
10615
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
0.00679275
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1789035718-uausAujY8HAmkQnztdDr"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1789035718-p9uUy53M0sj0IOsudl8n"
upstream_id
"edfe6286-b010-438b-a402-3e8bbb615234"
provider_responses
0
endpoint_id
"fd104bc8-535a-4d31-a847-a8e402e86a02"
id
"edfe6286-b010-438b-a402-3e8bbb615234"
is_byok
false
latency
340
model_permaslug
"deepseek/deepseek-v4.1-flash-20260910"
provider_name
"DeepSeek"
status
200
total_cost
0.00679275
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
86.38%
Matches word count
n/a
neededClean
false
words
511
97.87%
Dialogue to Total Word Ratio
Ratio: 29.34%, Deviation: 0.66%
neededClean
false
wordsTotal
518
wordsDialogue
152
92.1246%