NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-20-01M5C6C79V5G9PC5TAMWMMXY36
Run Details
Model:
DeepSeek V4.1 Flash 1017 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
17.8s
Cost:
$0.001083
Provider:
InferenceNet
Evaluation summary
Score
Evaluator
Details Link
0.03%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0152%
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 flattened the van’s sun visor and peered through the windshield at the bouncy castle wobbling in the Hendersons’ front garden. “So we’re agreed. We go in as clowns, grab the presents, and exit before the magic man finishes his second rabbit.” “No, Bill, we are not agreed,” Molly said, snapping the lid of the glove compartment. “Clowns are the first people they search. Every parent at that party has seen a true-crime documentary about a clown.” “That’s why it’s perfect. Nobody suspects the obvious clown.” “Everybody suspects the obvious clown. Last time you wore a disguise, you got arrested by a mall Santa.” “That was a jurisdictional misunderstanding. Besides, I’ve refined the plan. We tell them we’re from the birthday entertainment agency. We bring a piñata. We leave with the piñata and all the presents inside it.” Molly stared at him. “You want to stuff forty presents into a donkey?” “It’s a llama. And yes. It’s called supply-chain innovation.” “The piñata is made of papier-mâché, Bill. It’ll split before we reach the gate.” “Then we use the birthday cake. Have you seen the size of that thing? It’s a three-tier fortress. We hollow out the middle tier, load it with gift cards, and roll it out on a catering trolley.” “And the children?” “They’ll be distracted by the cake. That’s the beauty of it. We’re not stealing from children. We’re stealing from the concept of children’s parties.” Molly rubbed her temples. “What about the parents? The ones with phones and suspicion and, in one case, a Rottweiler named Princess?” “Princess loves me. I’ve got sausages.” “Of course you do.” “Look, the party starts at two. The magician arrives at two fifteen. The parents gather near the prosecco table at two twenty. That gives us a seven-minute window to swap the presents for wrapped bricks.” “Wrapped bricks?” “It’s a classic. They don’t notice until the kids open them. By then we’re on the M25 eating a bag of party rings.” “That’s the worst plan I’ve ever heard,” Molly said. “And I once helped you steal a van full of flotation devices by pretending to be a lifeguard at a swimming gala.” “That worked.” “We sank.” “Only a little. This time, we’ve got a clown car.” “We’re in a Fiesta.” “A clown Fiesta.” Bill grinned. “Molly, trust me. We walk in with balloons, we walk out with presents. What could possibly go wrong?” “The magician could be an off-duty police officer.” “He’s called Marvin the Mysterious.” “That’s exactly what an undercover officer would call himself.” Bill checked his watch. “Two minutes. Put on the wig.” “I’m not wearing the wig.” “Then we’re not a team.” Molly sighed and pulled the rainbow wig from the glovebox. “Fine. But if a six-year-old asks me to make a balloon dog, you’re doing the talking.”
created_at
"2026-10-20T08:20:49.601Z"
10/20/2026, 8:20:49 AM
model
"deepseek/deepseek-v4.1-flash-20261017"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
282
moderation_latency
(null)
generation_time
17767
tokens_prompt
75
tokens_completion
2279
native_tokens_prompt
67
native_tokens_completion
2155
native_tokens_completion_images
(null)
native_tokens_reasoning
1457
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
0.00108286
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1792484449-CDNOzaeu2XNh6XruoCNP"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1792484449-JAw4sLzvRmLWPnMK4U6q"
upstream_id
"gv2_CCnu4B68VIcBcb7ZdHt5S"
provider_responses
0
endpoint_id
"4231aea9-73af-4a2f-bd14-660b1d8b13b2"
id
"gv2_CCnu4B68VIcBcb7ZdHt5S"
is_byok
false
latency
282
model_permaslug
"deepseek/deepseek-v4.1-flash-20261017"
provider_name
"InferenceNet"
status
200
total_cost
0.00108286
cache_discount
(null)
upstream_inference_cost
0
provider_name
"InferenceNet"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.03%
Matches word count
n/a
neededClean
false
words
470
0.00%
Dialogue to Total Word Ratio
Ratio: 88.91%, Deviation: 18.91%
neededClean
false
wordsTotal
478
wordsDialogue
425
0.0152%