NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-20-01M5C6FM985GVE9SZ2YJ6892NA
Run Details
Model:
DeepSeek V4.1 Flash 1017 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
37.2s
Cost:
$0.003443
Provider:
InferenceNet
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
48.23%
Dialogue to Total Word Ratio
(details)
24.1154%
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 unfolded the party map across the dashboard while Bill polished a juice box with a tactical wipe. Outside, a bouncy castle wheezed beneath a banner for Timmy’s sixth birthday, and three clowns argued near the face-painting booth. “The perimeter is clowns, cake, and Brenda,” Molly said. “Brenda guards the presents like a dragon with frosting issues.” “I’m worried about the swag bin,” Bill said. “Every gift goes by the bouncy castle. We lift the bin, we’re golden.” “We are not lifting a bin full of toy trucks in front of forty toddlers. Toddlers are unreliable witnesses with sticky hands.” “So we go in as entertainment. I do magic. You do the grab.” “Your magic is pulling a quarter from a six-year-old’s ear. That’s assault with change.” “Fine. I’ll be the dinosaur mascot. Nobody questions a dinosaur carrying presents.” “Nobody questions a dinosaur because they’re screaming. Brenda will tase you.” “Brenda doesn’t have a taser.” “Brenda has a whistle and a look. The look is worse.” Bill tapped the map and squinted at the bouncy castle, where a birthday banner flapped in the wind. “We need a distraction. Pin the tail on the donkey, but the donkey is accidentally on fire.” “No fire. Last time you said tiny fire, and the insurance company sent us a pamphlet on arson.” “That pamphlet was laminated. Very professional.” Molly sighed and checked the party schedule. “Clean, quiet, festive. I go in as the balloon artist. I make a poodle, a sword, and a diversionary octopus. You slip the presents into the unicorn piñata.” “The piñata is empty. It’s for hitting.” “Exactly. Nobody suspects the thing they’re about to beat with a stick.” Bill considered the parents clustered near the cake table, already trading passive-aggressive remarks about gluten. The dashboard was littered with evidence of previous jobs: a laminated arson pamphlet, three expired coupons for clown college, and a polaroid of Bill dressed as a traffic cone. Molly pulled the van’s sun visor down and watched a parent wrestle a juice box away from a toddler who had clearly been training for this exact conflict. The bouncy castle sagged, rebounded, and issued a squeak that sounded like a budget being stretched. Near the gate, Brenda folded her arms and surveyed the party with the calm authority of someone who had once ejected a magician for sawing a birthday cake in half. Bill checked his watch twice, then a third time, as if expecting the schedule to apologize. The plan was simple, festive, and only mildly illegal. That was the problem. Molly folded the map into a paper hat and placed it on Bill’s head, where it immediately sagged. It was not a good sign. Bill saluted anyway. Molly groaned. “What about them?” “Cake and small talk buy us a four-minute window after Happy Birthday and before the sugar crash.” “And the children?” “Children see everything but report nothing accurately. By the time they describe us, we’ll be a dragon and a talking balloon.” “I want to be the dragon.” “You’re the balloon.” “I’m not being a balloon. Balloons pop.” “Then don’t get caught.” Bill frowned at the piñata. “One more thing. What if the presents are wrapped?” Molly stared at him until he checked the map again. “Bill. We’re stealing presents from a children’s party. Wrapped is the least of our problems.”
created_at
"2026-10-20T08:22:41.198Z"
10/20/2026, 8:22:41 AM
model
"deepseek/deepseek-v4.1-flash-20261017"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
134
moderation_latency
(null)
generation_time
37179
tokens_prompt
75
tokens_completion
6901
native_tokens_prompt
67
native_tokens_completion
6876
native_tokens_completion_images
(null)
native_tokens_reasoning
6097
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
0.00344336
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1792484561-vV9O9bsOIxXgtwgnuyTl"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1792484561-pdDkfRfW1ZuRlJOtNWn1"
upstream_id
"gv2_GQQJvKmn6Jhlre1lPnDem"
provider_responses
0
endpoint_id
"4231aea9-73af-4a2f-bd14-660b1d8b13b2"
id
"gv2_GQQJvKmn6Jhlre1lPnDem"
is_byok
false
latency
133
model_permaslug
"deepseek/deepseek-v4.1-flash-20261017"
provider_name
"InferenceNet"
status
200
total_cost
0.00344336
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
0.00%
Matches word count
n/a
neededClean
false
words
558
48.23%
Dialogue to Total Word Ratio
Ratio: 53.82%, Deviation: 3.82%
neededClean
false
wordsTotal
563
wordsDialogue
303
24.1154%