NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2DFH0ZPE6XFKZNACVRGA6N
Run Details
Model:
DeepSeek V4 Pro
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
25.5s
Cost:
$0.000739
Provider:
DeepSeek
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 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
Molly tapped the crinkled invitation against her chin, leaving a faint smear of chocolate from her thumb. “Alright, walk me through the layout again. I want to visualize the target package.” Bill sighed, scratching a patch of sunburn on his bald head. “For the last time, it’s a stack of presents, not a diplomatic pouch. You don’t need to visualize it, you just need to know that it’s sitting under a unicorn piñata.” “Unicorn piñata. Good. That’s a landmark,” she said, refusing to be rushed. “What’s the structural integrity of the decorative gift bow on the biggest box?” “I didn’t perform a tensile strength test on the frippery, Molly. It’s a bow. It’s made of flimsy hope and glitter.” “So it’ll collapse under pressure. Got it. Do we have eyes on the birthday princess?” Bill pulled a phone from his pocket, squinting at a photo he’d snapped through a hedge. “Tiny human, approximately three feet tall, wearing a taffeta gown that makes her look like a hostile cupcake. Answers to ‘Sophia.’ She’s the primary mark. Her mother is the one who hid our van keys under a box of frozen haddock last month. This is punitive and profitable.” “Revenge is a dish best served with a side of felony larceny,” Molly nodded, pulling a sparkly pink cowgirl hat from her bag and putting it on with the grim focus of a marine donning a helmet. “How are you getting past the clown?” “The clown is a seventeen-year-old named Kevin who smells like clove cigarettes and existential dread. I’m not worried about Kevin. I’m taking on the persona of Bumpo the Baffling, a wandering magician whose final trick is making a fifty-inch plasma-screen television vanish from a pile of Paw Patrol playsets.” “You’re doing magic?” Molly said, her face falling. “I’m doing a disgraced party princess. We’ll clash. A party can’t contain two competing entertainers. It’ll create a charisma vacuum, and the kids will riot.” “We are not entertainers, we are thieves,” Bill hissed, tugging a sequined vest over his grubby t-shirt. A stray sequin pinged off and hit a lamp. “My magic is a diversion. Your princess is the extraction. You’re the one with physical access to the gift table. What’s your tragic flaw, Princess Cowgirl Sparkle-Pants?” “I’m Princess Elara of the Whispering Winds, and my tragic flaw is that I’ve lost my enchanted voice and can only communicate through the solemn gift of mime and an elaborate interpretive dance,” she said, striking a pose of profound sorrow. “This justifies me silently weeping near the presents while miming the emotional weight of consumerism. I’ll swipe the big ones into this glitter sack and mime that it’s a magical portal.” Bill stared at her for a long moment. “You’re going to mime a magical portal?” “It’s method. These kids need to believe in the transformative power of loss.” “Fine. The signal for trouble is me pulling a never-ending rainbow handkerchief from my sleeve. The signal for ‘abort mission, the cake is a lie and it’s filled with off-duty cops’ is me pulling a very finite, single, grey handkerchief that looks sad.” “What’s the signal for the clown has sniffed out our ruse and is trying to twist you into a balloon animal of regret?” she asked. “That’s just screaming. High-pitched, unmodulated screaming. Now help me strap this collapsible sack trolley to my thigh. We’ve got a four-year-old to fiscally devastate.”
created_at
"2026-04-25T13:32:48.201Z"
4/25/2026, 1:32:48 PM
model
"deepseek/deepseek-v4-pro-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1035
moderation_latency
(null)
generation_time
25197
tokens_prompt
77
tokens_completion
925
native_tokens_prompt
67
native_tokens_completion
816
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
0.000739065
router
(null)
provider_responses
0
endpoint_id
"d084170b-dd99-4e53-9ae1-b63881f3cfbf"
id
"aa2f0c3b-1ba6-446f-8a95-2ba1d6e72b3e"
is_byok
false
latency
351
model_permaslug
"deepseek/deepseek-v4-pro-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-1777123968-wnrL0WaALEUdtDqeeiwA"
session_id
(null)
api_type
"completions"
id
"gen-1777123968-niow6OME69hEDl0Y64yB"
upstream_id
"aa2f0c3b-1ba6-446f-8a95-2ba1d6e72b3e"
total_cost
0.000739065
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
569
0.00%
Dialogue to Total Word Ratio
Ratio: 79.79%, Deviation: 29.79%
neededClean
false
wordsTotal
579
wordsDialogue
462
0.0000%