NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5TXTRFYYAAMJ3VTK1C1X3
Run Details
Model:
DeepSeek V3.2
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
41.2s
Cost:
$0.000367
Provider:
Novita
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 air in the rented storage unit smelled of rust and defeat. Molly, her face a map of skeptical lines, stared at the blueprint taped to a stack of moldering chairs. It was a diagram of a suburban backyard, meticulously annotated. “Let me get this straight,” Molly said, not looking at Bill. “We’re skipping the diamond exchange, the museum gala, the entire *city*, to hit a… sixth birthday party for a kid named Liam.” Bill, polishing a pair of miniature bolt cutters with theatrical care, beamed. “Not just any party, Moll. A ‘Super-Spectacular Mega-Superhero Extravaganza.’ It’s a target-rich environment. Thirty kids, each with a present. Low security, predictable patterns, and an exit strategy that involves a piñata distraction.” “Our fence is going to give us fifty cents on the dollar for a pile of action figures and glitter slime.” “Ah, but you’re thinking small!” Bill produced a tablet, swiping to a grainy photo of a lavish gift table. “See that silver-wrapped box, slightly larger? That’s from Great-Aunt Edna. She’s old money and buys boring, sensible gifts. Like gold bonds. Or a solid silver rattle from 1922.” “He’s six. She’s buying him a book about trains.” Bill swiped again. “And this oblong shape? From the tech-uncle. That’s not a toy lightsaber. That’s a high-end drone. The retail on that could fund three months of our ‘avoiding the landlord’ fund.” Molly leaned in, her professional curiosity reluctantly piqued. “Security?” “Two-tiered. Outer perimeter: a gaggle of sleep-deprived parents, focused primarily on preventing cake-induced meltdowns. Inner sanctum: one harassed mother-of-the-birth-boy, likely armed with baby wipes and a desperate need for pinot grigio. The real wildcard is the entertainment.” “Clown?” “Worse. Captain Valiant. A local theatre student in a sweaty nylon suit. His powers include mediocre balloon animals and a shaky grasp of childhood consent regarding high-fives.” Molly paced, her boots echoing on the concrete. “So we infiltrate as…?” “Party caterers. Specifically, the specialty ‘Hero Juice’ punch bowl attendants. I’ve got the aprons. We arrive early, case the layout, and during the piñata frenzy—a peak chaos event of astonishing violence—we switch the gift table with an identical, pre-staged table of empty boxes.” “In broad daylight.” “Under a canopy of Spider-Man tablecloths! It’s audacious. It’s so stupid, it’s brilliant.” Molly stopped, pinching the bridge of her nose. “Bill. The loot. We’re two professionals debating the best way to steal a plush Hulk. We have standards.” “Do we?” Bill asked, his voice dropping to a conspiratorial whisper. “Remember the Basel job? Laser grids, pressure plates, a six-month planning phase? We got nabbed by a janial on a smoke break. This… this is simplicity. This is humility. This is, frankly, a cry for help, but also a guaranteed score.” A long silence hung in the dusty air. Molly looked at the blueprint, at the tiny drawing of a bouncy castle labeled ‘primary diversion.’ “The piñata,” she said finally. “It’s our window. But we need a fallback. If Captain Valiant gets suspicious…” Bill grinned, pulling a small, glitter-filled pouch from his pocket. “I’ve pre-loaded a water pistol with this. A squirt to the face, and he’ll be too busy looking like a disco ball to see a thing. We’ll be in the van, debating the secondary market for light-up sneakers, before the first tear is shed over shattered candy dreams.” Molly sighed, a sound of utter resignation. “Fine. But if I have to make small-talk about pediatric dentistry with a man dressed as a flag, I’m keeping the drone.”
created_at
"2026-02-16T21:32:03.776057+00:00"
2/16/2026, 9:32:03 PM
model
"deepseek/deepseek-v3.2-20251201"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1586
moderation_latency
(null)
generation_time
31051
tokens_prompt
77
tokens_completion
997
native_tokens_prompt
67
native_tokens_completion
873
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
0.000367223
router
(null)
provider_responses
0
status
(null)
is_byok
false
latency
10077
endpoint_id
"b39a588a-b338-43ba-b046-226b9e24afa8"
provider_name
"SiliconFlow"
model_permaslug
"deepseek/deepseek-v3.2-20251201"
1
id
"4cdd73233552741ffe191e71d4a5a7ef"
status
200
is_byok
false
latency
1584
endpoint_id
"8a5fb6d3-ef71-4958-af54-269ffa3a3c6e"
provider_name
"Novita"
model_permaslug
"deepseek/deepseek-v3.2-20251201"
api_type
"completions"
id
"gen-1771277481-6S3SY4CxWAugcKu4aA1g"
upstream_id
"4cdd73233552741ffe191e71d4a5a7ef"
total_cost
0.000367223
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Novita"
Evaluation details
Result
Evaluator
Details
Meta Data
0.0000%
Matches word count
n/a
neededClean
false
words
600
0.0000%
Dialogue to Total Word Ratio
Ratio: 74.67%, Deviation: 44.67%
neededClean
false
wordsTotal
600
wordsDialogue
448
0.0000%