NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-11-01KKFDEXJRF1WPNAGHFWGZF1HK
Run Details
Model:
Stealth: Hunter Alpha
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
19.8s
Cost:
$0.000000
Provider:
Stealth
Evaluation summary
Score
Evaluator
Details Link
43.38%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
21.6892%
Overall
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 chalk outline of a house on the warehouse floor was not a crime scene; it was a tactical blueprint. Molly crouched beside it, a laser pointer dancing across a crudely drawn bouncy castle. “Phase one,” she said, her voice tight with manufactured drama. “Infiltration. We go in as the entertainment. You’re Sparkles the Clown.” Bill, who was meticulously braiding a lockpick into a jester’s wig, looked up. “I hate clowns. They’re objectively terrifying. How about a pirate? Pirates are cool.” “Pirates don’t hand out balloons, Bill. A clown provides plausible cover for loitering near the gift table. Besides,” she added, studying a stolen invitation, “the birthday boy, Timmy, is terrified of parrots. A pirate would cause a scene.” “Fine,” he sighed, testing the spring of a whoopee cushion. “But I get to keep the nose. It’s got a lovely squeak.” “Phase two. Distraction.” Molly clicked the laser to a new spot, illuminating a drawing of a magician’s top hat. “I’m Madame Zamboni, the mystic. While I’m telling fortunes about homework and lost hamsters, you unleash the swarm.” “Right, the swarm.” Bill patted a duffel bag that buzzed faintly. “A hundred and twelve robotic ladybugs, each programmed to seek out and lift shiny objects. Untraceable. Adorable.” “They’re not adorable, Bill. They’re a technological marvel. Once the gifts are tagged, we proceed to phase three: the exit. During the piñata chaos—” “Which we’ll manufacture with a strategically swung baseball bat,” Bill interjected. “—we merge with the departing parent herd. The ladybugs deposit the loot in the trunk of our ‘Mom Van,’ we change in the back, and drive away discussing soccer practice. It’s airtight.” Bill stood, stretching. “There’s one flaw in your beautiful, beautiful plan, Mols.” “Which is?” “The presents. We’re stealing… let’s see, from this guest list… a pirate ship playset, a dinosaur dig kit, and about forty pounds of assorted plastic rubbish.” He held up a hand as she started to protest. “I’m in, obviously. But the take is less ‘priceless jewels’ and more ‘what’s on sale at Mega-Mart.’” Molly grinned, a flash of pure, undiluted greed. “Bill, you’re thinking like a thief, not a visionary. What do desperate, sleep-deprived parents want more than anything?” “A nap?” “They want *quiet*. And I’ve secured a buyer—a consortium of suburban dads—for the entire haul. They don’t want the toys. They want the warranty cards. The instruction manuals. The *leverage*.” Bill’s eyes widened. “You mean…?” “Exactly. No pirate ship can be assembled without page six of the instructions. No robot dinosaur can stomp without its proprietary battery pack. We’re not stealing toys, my friend. We’re stealing peace. And for that, they’ll pay a premium.” She snapped the laser off. “Now, try on the nose. We need to perfect the squeak for comedic timing.” The warehouse fell silent, save for the pathetic, hopeful *eep* of Bill’s clown nose being squeezed, over and over again.
created_at
"2026-03-11T21:39:48.451Z"
3/11/2026, 9:39:48 PM
model
"openrouter/hunter-alpha"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1660
moderation_latency
(null)
generation_time
19773
tokens_prompt
77
tokens_completion
1158
native_tokens_prompt
2111
native_tokens_completion
974
native_tokens_completion_images
(null)
native_tokens_reasoning
320
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-1773265188-sVlhhbHOflIrjv9rZ9ko"
upstream_id
"gen-1773265188-sVlhhbHOflIrjv9rZ9ko"
total_cost
0
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Stealth"
Evaluation details
Result
Evaluator
Details
Meta Data
43.38%
Matches word count
n/a
neededClean
false
words
483
0.00%
Dialogue to Total Word Ratio
Ratio: 70.10%, Deviation: 40.10%
neededClean
false
wordsTotal
485
wordsDialogue
340
21.6892%