NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5N5MJ4H7CHBKPTCXGDJ2F
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
26.5s
Cost:
$0.000740
Provider:
DeepInfra
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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
The dim glow of a flickering streetlamp illuminated the cracked pavement outside Molly and Bill’s hideout, a dingy apartment above a laundromat that smelled perpetually of detergent and regret. Inside, the pair hunched over a crumpled piece of paper on the coffee table, their faces illuminated by the faint light of a single bulb swinging precariously from the ceiling. The paper was a rough sketch of a suburban house, complete with stick-figure children and a disproportionate dog labeled “Cerberus.” Molly tapped the paper with a chipped fingernail. “This is the target. Birthday party for a six-year-old named Timmy. According to my intel, the living room is *littered* with presents. Toys, gadgets, the works. It’s a gold mine.” Bill frowned, scratching the stubble on his chin. “A kid’s party? Molly, we’re thieves, not monsters.” Molly rolled her eyes. “Oh, come on. They’re kids. They’ll get over it. Besides, it’s not like we’re stealing their birthdays. We’re just… redistributing their presents. Think of it as a public service.” Bill sighed, rubbing his temples. “I don’t know. It feels… wrong. What if it’s his bike? His favorite teddy bear? His—” “His brand-new PlayStation Five?” Molly interrupted, raising an eyebrow. “Which, by the way, we can sell for a fortune. Focus, Bill. This is a low-risk, high-reward gig. No alarms, no guards, just a bunch of sugar-crazed kids and distracted parents.” Bill hesitated, glancing at the sketch again. “What about the dog? You labeled it ‘Cerberus.’ That doesn’t sound good.” Molly waved her hand dismissively. “It’s a Chihuahua. Barking powder will handle it. I’ve got a fresh batch.” Bill groaned, sinking into the worn couch. “Barking powder? Last time you used that, we spent three hours cleaning up glitter.” Molly grinned. “Details, details. Look, here’s the plan: I’ll go in through the backyard, posing as a clown. You’ll sneak in through the side window during the cake chaos. We grab the loot, meet at the getaway car, and we’re home free.” Bill stared at her, incredulous. “You’re going as a clown?” She shrugged. “What better disguise? Nobody questions a clown at a kid’s party. I’ve got the costume ready. Big shoes, red nose, the whole shebang.” Bill groaned again. “I hate clowns.” “Nobody likes clowns, Bill,” Molly said, patting his shoulder. “That’s why it’s perfect.” He sighed, resigned. “Fine. But if anything goes wrong—” “Nothing’s going to go wrong,” Molly interrupted, grabbing her oversized polka-dotted clown bag from the corner. “Trust me. This is going to be the easiest score of our careers.” Bill muttered under his breath, “Famous last words,” as he followed her out the door into the chilly night. The drive to the neighborhood was uneventful, the streets eerily quiet as they approached Timmy’s house. Lights blazed from the windows, and the sound of children’s laughter echoed through the air. Molly parked the car a few houses down, turning to Bill with a grin. “Showtime.” Bill groaned one last time as they stepped out of the car, Molly adjusting her oversized red nose and Bill tugging at the collar of his inconspicuous black hoodie. The plan was set, but as they crept toward the house, Bill couldn’t shake the sinking feeling that something was about to go terribly, hilariously wrong.
created_at
"2026-02-16T21:28:40.046572+00:00"
2/16/2026, 9:28:40 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
352
moderation_latency
(null)
generation_time
26431
tokens_prompt
77
tokens_completion
909
native_tokens_prompt
66
native_tokens_completion
808
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
0.00074024
router
(null)
provider_responses
0
id
"chatcmpl-R8Xz2scvY7DttxyVoptd2cvy"
status
200
is_byok
false
latency
60
endpoint_id
"5294d55f-9012-496b-8f22-8cc919432dcd"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277293-LWbeiiQ75R3yjcVnD3gj"
upstream_id
"chatcmpl-R8Xz2scvY7DttxyVoptd2cvy"
total_cost
0.00074024
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
0.0000%
Matches word count
n/a
neededClean
false
words
547
0.0000%
Dialogue to Total Word Ratio
Ratio: 49.54%, Deviation: 19.54%
neededClean
false
wordsTotal
547
wordsDialogue
271
0.0000%