NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-10-01M25DR4N1P4WWK3K96DR6TV3Z
Run Details
Model:
DeepSeek V4.1 Flash (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
20.3s
Cost:
$0.003105
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
49.9595%
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
"Bill, walk me through the plan again," Molly said, adjusting the tiny party hat she'd stolen from a clown. "Because last time you said 'trust me' and I ended up in a bouncy castle with a warrant." "That was a tactical bounce," Bill said. "This is simpler. We infiltrate the Wigglesworth twins' seventh birthday, locate the present pile, and extract every wrapped box before Pin the Tail on the Donkey concludes." "Extract how? There are forty parents, a magician, and a petting zoo." "The magician's a distraction. I hired him." "You hired a magician for a heist?" "I hired him to be terrible. He'll make a rabbit disappear and then cry. Nobody watches presents when a grown man is weeping into a top hat." Molly pinched the bridge of her nose. "And the petting zoo?" "Goats are indifferent to crime. I checked." "You checked?" "I asked one. He ate my notebook. That's a yes." "Okay. Entry?" "Back gate. I've got a clipboard that says I'm from Party Compliance." "Party Compliance isn't a thing." "It is now. I laminated it." "Of course you did. What about the kids?" "Kids are easy. They're small, loud, and easily bribed with sugar. We toss a handful of jelly beans into the ball pit and they swarm like piranhas in party hats." "That's monstrous." "It's efficient. Besides, we're not stealing from kids. We're stealing from the concept of kids' birthday parties. Which is, frankly, a public service." "The concept has a receipt. And a registry." "Which we will also steal." Molly sighed. "Let's suppose we get to the presents. There's a mountain of them." "Pyramid. I've seen the photos. Thirty-seven presents. Two bicycles, one drone, a chemistry set, and what I believe is a pony." "A pony isn't a present." "It's in a bow. It's a present." "How do we get a pony out the back gate?" "Very quietly. And with carrots." "Bill." "Fine, we skip the pony. But we take the drone." "The drone is for a seven-year-old." "And now it's for a forty-two-year-old criminal mastermind." "You're thirty-six." "I rounded up. Dignity." Molly leaned in. "What about the clown?" "Which clown?" "The one who's been staring at us from the shrubbery." Bill turned slowly. "That's not a clown. That's my ex-wife." "Is she a problem?" "Only if she recognizes me. And if she recognizes you." "Why would she recognize me?" "Because you're wearing her party hat." Molly removed the hat. "New plan. We go in, we take the presents, we blame the magician." "Now you're thinking like a professional." "And we leave the pony." "Agreed. Ponies have lawyers." "Wait," Molly said. "What about the cake?" "We leave it. It's probably gluten-free." "You're a monster." "A monster with a drone." "And a clipboard," Bill said. "Don't forget the clipboard." "Party Compliance," Molly muttered. "I hate this job." They shook hands. Then the clown in the shrubbery coughed and said, "I heard everything. I want thirty percent." Bill looked at Molly. "See? This is why I hate children's parties."
created_at
"2026-09-10T10:27:53.638Z"
9/10/2026, 10:27:53 AM
model
"deepseek/deepseek-v4.1-flash-20260910"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
921
moderation_latency
(null)
generation_time
20214
tokens_prompt
75
tokens_completion
4461
native_tokens_prompt
93
native_tokens_completion
5151
native_tokens_completion_images
(null)
native_tokens_reasoning
4418
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
0.00310455
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1789036073-x0CFEQ4CsBzVWDhwP0WS"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1789036073-yc8ofBKsM3kSaF9rMaEt"
upstream_id
"3aa68bea-c093-4a92-8341-b418baaafebd"
provider_responses
0
endpoint_id
"fd104bc8-535a-4d31-a847-a8e402e86a02"
id
"3aa68bea-c093-4a92-8341-b418baaafebd"
is_byok
false
latency
347
model_permaslug
"deepseek/deepseek-v4.1-flash-20260910"
provider_name
"DeepSeek"
status
200
total_cost
0.00310455
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
99.92%
Matches word count
n/a
neededClean
false
words
503
0.00%
Dialogue to Total Word Ratio
Ratio: 89.24%, Deviation: 19.24%
neededClean
false
wordsTotal
511
wordsDialogue
456
49.9595%