NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-11-01KKFDCAFE5S0BE0GBC69AHFEW
Run Details
Model:
Stealth: Hunter Alpha
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
26.6s
Cost:
$0.000000
Provider:
Stealth
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
10.84%
Dialogue to Total Word Ratio
(details)
5.4184%
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
The plastic tablecloth rippled in the breeze, a pastel banner proclaiming "Happy 6th Birthday, Toby!" sagged between two oak trees, and the cacophony of thirty sugar-fueled children formed a perfect sonic smokescreen. Molly lowered her binoculars. "The demographic is our greatest asset. Unpredictable, loud, and easily bribed with glitter." Bill chewed his lip, eyeing the inflatable castle. "I don't know, Moll. This feels… ethically squishy. We're talking about a piñata's worth of loot here." "Ethically squishy is our business model," Molly said, unfolding a blueprint on the hood of their nondescript minivan. It was a detailed map of the suburban backyard, with dotted lines labeled 'Chaos Corridors' and 'Tantrum Zones.' "Phase One: Distraction. You deploy the Super-Soaker Squadron at the lemonade stand. Create a perimeter of wet, sugary chaos." "They're kids, Bill. Their situational awareness is nil. They're currently engaged in a mass screaming contest. Phase Two: I, disguised as the party clown—" "Absolutely not." "—insert myself into the gift table perimeter. The presents are pre-sorted by parental value. The big, rectangular ones from 'Aunt Susan' are the primary targets. We leave the stuffed animals. That's our moral compass." Bill sighed, a long, weary sound. "Our moral compass is pointing directly at a six-year-old's new bike." "Correction: a six-year-old's *future* bike. He hasn't formed an attachment yet. It's pre-nostalgia. Phase Three: the switch." She patted a large, unmarked cardboard box in the van's back seat. "We replace the loot with identically wrapped boxes containing… educational workbooks." Bill stared at her. "You're a monster." "I'm a visionary. The parents will be thrilled. Toby will learn long division. It's a public service. Now, are you in, or are you going to let your conscience be troubled by a bunch of second-graders high on cake?" Bill watched as a small boy in a cape attempted to fly off a garden wall. "What about security?" "Look at them. The only guard is a golden retriever asleep under the buffet table. The only weapon is a water balloon launcher manned by a kid who can't aim. We're not robbing Fort Knox, Bill. We're liberating consumer goods from a temporary tyranny of joy." He finally nodded, a slow capitulation. "Fine. But I'm not wearing a clown nose." "Deal. You get the Super-Soakers. Remember, aim for the shoes. It causes maximum disarray with minimal actual harm. And for heaven's sake, try to look like a fun uncle." Five minutes later, Molly, in a rainbow wig and tragically painted smile, was weaving between screaming children, her oversized shoes squeaking on the grass. Bill, armed with two neon-green water cannons, created a diversionary tidal wave near the snack table. As planned, parental attention snapped toward the soggy crackers and crying toddlers. Molly reached the gift table, a mountain of brightly colored packages. With practiced speed, she began the swap, her clown gloves making the wrapping paper rustle. The big, rectangular 'Aunt Susan' present vanished into her oversized prop bag, replaced by the workbook-filled decoy. "Clown! Clown, do a trick!" a small girl demanded, tugging at her polka-dotted pants. Molly froze. This wasn't in the blueprint. She fumbled in her pocket and produced a handful of rubber chickens, squeaking them frantically. The girl shrieked with delight, drawing more children. A circle formed. They demanded more. Across the yard, Bill was losing his battle. He’d been disarmed by a seven-year-old general and was now being pelted with his own ammunition. He caught Molly’s eye over the heads of the mob and gave a panicked, water-logged shrug. Molly squeaked the chickens one last time, executed a clumsy bow, and began backing toward the van, a trail of children and honking poultry in her wake. It was not the clean, silent exit she had envisioned. But as she tossed the bag of presents into the back and peeled out of the cul-de-sac, Bill wringing water from his shirt beside her, she allowed herself a small, red-lipped smile. "The long division workbooks were a nice touch," Bill conceded, watching the birthday chaos recede in the side mirror. "Everyone learns something today," Molly said, adjusting her rainbow wig. "Especially Toby. He's about to learn the fundamental truth of the universe: you can't always get what you want."
created_at
"2026-03-11T21:38:23.354Z"
3/11/2026, 9:38:23 PM
model
"openrouter/hunter-alpha"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3788
moderation_latency
(null)
generation_time
26495
tokens_prompt
77
tokens_completion
1407
native_tokens_prompt
2111
native_tokens_completion
1243
native_tokens_completion_images
(null)
native_tokens_reasoning
271
native_tokens_cached
1984
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
router
(null)
provider_responses
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.0; linux; x64))"
http_referer
(null)
api_type
"completions"
id
"gen-1773265103-ParV3a2tiMJuajVCheDG"
upstream_id
"gen-1773265103-ParV3a2tiMJuajVCheDG"
total_cost
0
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Stealth"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
703
10.84%
Dialogue to Total Word Ratio
Ratio: 43.33%, Deviation: 6.67%
neededClean
false
wordsTotal
720
wordsDialogue
312
5.4184%