NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2G8PB7J04DYDR0F7ZQ2W5R
Run Details
Model:
DeepSeek V4 Flash
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
11.9s
Cost:
$0.000225
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
9.61%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
4.8041%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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
“I’m telling you, Bill, it’s a goldmine.” Molly jabbed a finger at the crumpled flyer on the table. “Thirteen seven-year-olds. Balloons. A bouncy castle. And a mountain of wrapped treasure.” Bill squinted at the ink. “You mean presents.” “Presents are treasure when you’re seven. And their parents are rich.” Molly leaned back in the creaky diner booth. “The theme is unicorns. There’s a piñata shaped like a rainbow.” “I’m not hitting a piñata with a seven-year-old.” “You won’t have to. You’re the distraction.” Bill rubbed his temples. “Distraction. At a unicorn party. What’s my cover? Sparkle clown? Magical llama handler? Because I draw the line at a unitard.” “No unitard,” Molly said. “You’re the balloon guy.” “I don’t do balloon animals. I tried once. Made a poodle that looked like a scrotum.” “You’re not making animals. You’re making swords.” “Swords. For children.” “Swords pay, Bill. They’re easy. Two twists, a fold. They love them. While you’re handing out balloon weaponry, I slip through the back gate, hit the gift table, and we’re gone before the parents realize the piñata’s been looted.” Bill picked at a fry. “And how do I explain the tactical vest under my polo shirt?” “You don’t. You’re the quirky uncle. The weird one who smells like menthol and regret. Kids love that.” “I don’t smell like regret. I smell like cheap cologne and low self-esteem.” “Same diff. Look, the real challenge is the cake. It’s a unicorn cake. Six feet tall. Fondant horn. If you’re not careful, you’ll knock it over and cause a sugary apocalypse.” “Who knocks over a cake on purpose?” “Not on purpose. As a distraction. You bump into it, the kids scream, I grab the haul, we meet at the car.” Bill stared at her. “So my role in this caper is: arrive early, make bad balloon swords, then commit cake-homicide in front of a dozen weeping children.” “It’s a clean job. No violence. Just tears.” “Their tears, or mine?” “Mostly theirs. Yours will be from the shame, which is temporary. I’ve timed the sprinkler system in the backyard. If we’re fast, the parents will blame the crying on the wet grass.” “And if I get caught?” “You’re the balloon guy. You’re a mystery. No one remembers the balloon guy’s face. They remember the crotch-shaped poodle.” Bill sighed, long and deep. “Fine. But I’m not doing the cake. Find another patsy.” “You’re not a patsy. You’re the keystone. The lynchpin. The balletic chaos agent.” “I’m a thief with a latex allergy and a weak knee.” “Perfect. You’ll fit right in with the other broken dads.” Molly slid out of the booth. “I’ll text you the address. Party’s at three. Wear your vest under a cardigan.” “I don’t own a cardigan.” “Buy one. And Bill?” “What?” “Don’t eat the fondant. It’s for the kids. And we’re monsters, but we’re not animals.”
created_at
"2026-04-25T14:21:29.835Z"
4/25/2026, 2:21:29 PM
model
"deepseek/deepseek-v4-flash-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
524
moderation_latency
(null)
generation_time
11790
tokens_prompt
77
tokens_completion
832
native_tokens_prompt
67
native_tokens_completion
771
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)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.00022526
router
(null)
provider_responses
0
endpoint_id
"722e0746-2034-4d6e-a08f-e6271849ee9e"
id
"81ed052d-3cef-46a6-a801-7197f3de681c"
is_byok
false
latency
272
model_permaslug
"deepseek/deepseek-v4-flash-20260423"
provider_name
"DeepSeek"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777126889-M7mna2yT7xgHDDhbpqFW"
session_id
(null)
api_type
"completions"
id
"gen-1777126889-7ZkP8RuWWq4DpDoLLkW2"
upstream_id
"81ed052d-3cef-46a6-a801-7197f3de681c"
total_cost
0.00022526
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
9.61%
Matches word count
n/a
neededClean
false
words
478
0.00%
Dialogue to Total Word Ratio
Ratio: 89.69%, Deviation: 19.69%
neededClean
false
wordsTotal
485
wordsDialogue
435
4.8041%