NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2H8J170Q4CK40RM8BE3CZ8
Run Details
Model:
DeepSeek V4 Pro
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
10.5s
Cost:
$0.000286
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
81.27%
Matches word count
(details)
36.79%
Dialogue to Total Word Ratio
(details)
59.0303%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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, the cake is a lie." "I swear to you, Bill, the cake is a seven-tier unicorn monstrosity full of cash." "Explain the cash part again." "The birthday boy is little Timmy Vanderplume. His dad owns a chain of dry cleaners that’s clearly a front. Timmy’s gifts are just hollowed-out board books stuffed with hundreds." "So we’re crawling through a bouncy castle air shaft?" "It’s the only way. The piñata blindfold system is too unreliable." "What’s our cover?" "You’re Crusty the Clown." "Absolutely not. Clowns have dead eyes. I’ll be a magician." "Fine. Your magic trick is making a Nintendo Switch disappear into your oversized trousers." "Brilliant. What about the security?" "Kindergarteners with juice-box sticky hands. Lethal in close quarters. We’ll need a diversion." "Like what?" "I hired a seven-foot man in a Bluey costume." "Standard misdirection. He dances the floss, the kids swarm him, parental sightlines break, the bouncy castle becomes a void of chaos, and we load the VTech smartwatches." "What’s our exit?" "The ice cream van, obviously. The chime covers all auditory evidence." "Molly, I ask you this with complete sincerity: what flavor?" "Larceny ripple."
created_at
"2026-04-25T14:38:53.996Z"
4/25/2026, 2:38:53 PM
model
"deepseek/deepseek-v4-pro-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1247
moderation_latency
(null)
generation_time
10444
tokens_prompt
77
tokens_completion
317
native_tokens_prompt
67
native_tokens_completion
295
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
0.000285795
router
(null)
provider_responses
0
endpoint_id
"d084170b-dd99-4e53-9ae1-b63881f3cfbf"
id
"e5d3aa1d-21b5-4c71-bec3-4121c6e1e1cb"
is_byok
false
latency
278
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-1777127933-q2NHPposiKtBNmpitFn8"
session_id
(null)
api_type
"completions"
id
"gen-1777127933-fIQsjDHYX6wSB09zOAg8"
upstream_id
"e5d3aa1d-21b5-4c71-bec3-4121c6e1e1cb"
total_cost
0.000285795
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
81.27%
Matches word count
n/a
neededClean
false
words
188
36.79%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 10.00%
neededClean
false
wordsTotal
192
wordsDialogue
192
59.0303%