NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5YMCJHRYF69ATEWKD3WZA
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
29.4s
Cost:
$0.000723
Provider:
DeepInfra
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
Molly leaned back in her chair, twirling a strand of her hair around her finger. “So, let me get this straight. You want to break into a kid’s birthday party?” Bill leaned forward, his eyes gleaming with excitement. “Not just any kid’s party. Bobby Carmichael’s. His dad’s that tech billionaire. You think the presents are gonna be anything less than gold-plated?” Molly raised an eyebrow. “And what, we’re just gonna waltz in there dressed as clowns and grab the loot?” Bill grinned. “Clowns? Nah, too obvious. We’re gonna go as magicians.” Molly snorted. “Magicians? You don’t even know how to shuffle a deck of cards.” “Details, Molly, details,” Bill said, waving a hand dismissively. “We’ll buy one of those magic kits from the toy store. Presto-change-o, we’re in.” Molly rolled her eyes. “And how exactly are we gonna get the presents out? Stuff ‘em in your oversized hat?” Bill tapped his temple. “Already thought of that. We’ll have a getaway van parked behind the house. We’ll distract the kids with a ‘magic show,’ whisk the presents away ‘stage left,’ and boom—we’re outta there.” Molly stared at him for a long moment. “You do realize kids are like bloodhounds, right? They’ll notice if their presents suddenly vanish.” “Not if we replace ‘em with empty boxes,” Bill said smugly. “We’ll say it’s part of the trick.” Molly sighed. “Bill, this plan has more holes than a slice of Swiss cheese.” Bill leaned back, crossing his arms. “Got a better idea?” Molly smirked. “Actually, yeah. We go as caterers.” Bill frowned. “Caterers?” “Yeah,” Molly said, leaning in. “We show up with a bunch of trays of cupcakes. Nobody questions caterers. We blend in, slip the presents into the cupcake boxes, and walk out like we’re just packing up.” Bill scratched his chin. “Huh. Not bad. But what about the parents? They’re not gonna just let us walk out with all those presents.” Molly shrugged. “We’ll say they’re party favors. Rich people love that kind of crap.” Bill grinned. “You know, Molly, you might just be a genius.” “Might be?” Molly shot back with a smirk. “Try ‘definitely am.’” Bill laughed. “Alright, genius, what’s the next step?” Molly tapped her fingers on the table. “We scope out the house. Figure out the layout, where they’re keeping the presents, all that jazz.” Bill nodded. “And when do we hit the party?” Molly checked her phone. “Bobby’s birthday is next Saturday. We’ve got a week to prep.” Bill rubbed his hands together. “Perfect. Operation Birthday Bash is a go.” Molly rolled her eyes. “Operation Birthday Bash? Really?” “Got a better name?” “How about Operation Don’t Get Caught?” Bill chuckled. “Touché.” Molly stood up, grabbing her jacket. “Alright, let’s get to work. And Bill?” “Yeah?” “Try not to screw this up.” Bill placed a hand over his heart, feigning offense. “Molly, when have I ever let you down?” Molly paused, giving him a deadpan look. “Do you really want me to answer that?” Bill chuckled nervously. “Fair point. Let’s just stick to the plan.” Molly smirked. “For once, I actually agree with you.”
created_at
"2026-02-16T21:33:53.064371+00:00"
2/16/2026, 9:33:53 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
246
moderation_latency
(null)
generation_time
29322
tokens_prompt
77
tokens_completion
907
native_tokens_prompt
66
native_tokens_completion
789
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)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.00072333
router
(null)
provider_responses
0
id
"chatcmpl-R0JvCg39H3zl9KCNKjLNaETl"
status
200
is_byok
false
latency
65
endpoint_id
"5294d55f-9012-496b-8f22-8cc919432dcd"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277603-TA70O4IQXJ7pCqcOh8OI"
upstream_id
"chatcmpl-R0JvCg39H3zl9KCNKjLNaETl"
total_cost
0.00072333
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
20.1897%
Matches word count
n/a
neededClean
false
words
520
58.6025%
Dialogue to Total Word Ratio
Ratio: 73.27%, Deviation: 3.27%
neededClean
false
wordsTotal
520
wordsDialogue
381
39.3961%