NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHY98SSW3G87NZK72N4C2TK8
Run Details
Model:
Gemma 3 12B
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
13.9s
Cost:
$0.000092
Provider:
DeepInfra
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 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 community hall smelled overwhelmingly of sugar and desperation. Balloons bobbed like panicked goldfish, and a banner proclaiming "Barnaby's 5th Birthday Bash!" drooped over a table laden with plastic cutlery and suspiciously bright orange juice. Molly surveyed the scene with a professional eye, adjusting the oversized sunglasses perched on her nose. “Right, Bill,” she said, her voice low, “Operation Sugar Rush is a go. Observe the parameters. We're targeting the present mountain. Priority: Lego. Secondary: anything with batteries. Tertiary: that enormous fluffy unicorn. It's hideous, but incredibly valuable on the collector’s circuit." Bill, a man whose enthusiasm often outstripped his competence, bounced on the balls of his feet. He was currently disguised as a particularly anxious-looking clown, complete with a wig that threatened to achieve escape velocity and a nose that glowed faintly in the dim light. “Lego, batteries, unicorn… Roger that, Molly! But… are we *sure* about the unicorn? It looks… sad.” Molly sighed, pinching the bridge of her nose. "Bill, we're stealing presents, not offering therapy to plush livestock. Focus. Security is minimal. One bored-looking security guard playing Candy Crush in the corner, a gaggle of hyperactive five-year-olds, and a parent or two desperately clinging to sanity. The weakest link is clearly the cake table, deceptively close to the present pile. We create a distraction, I grab the loot, you… well, you try to look less like you're about to burst into tears." “I can create a distraction! I’m a clown! I can juggle! I can… make balloon animals!” Bill brightened, attempting to twist a balloon into what vaguely resembled a dachshund. It promptly exploded. Molly ignored the popping sound. “No juggling, Bill. No balloon animals. We're aiming for plausible deniability, not a circus. I want you to stage an elaborate, but entirely harmless, meltdown. Dramatic fainting, maybe? Claim an allergy to streamers? Just… something to draw attention away from the present mountain while I work.” “A meltdown, you say? Can I practice? I think I could really sell a despairing performance about… about losing my red nose!” Bill wrung his hands, his clown makeup smeared slightly. “Just don’t actually lose your nose, Bill. And for the love of all that is holy, no crying. We are professionals.” Molly pointed toward the security guard. "He’s our key. I’ll handle him. Tell him his Candy Crush streak is about to be disrupted by a rogue clown. Something like that. Just… keep it vague and dramatic. And make sure he looks thoroughly annoyed." “Right. Annoyed. Vague. Meltdown. Unicorn. Got it!" Bill took a deep breath, attempting a mournful expression. "Oh, woe is me! The joys of childhood… fleeting… like a rainbow… a slightly deflated rainbow…” Molly winced. "Just… try to be convincingly distressed, Bill. We have precisely fifteen minutes before Barnaby starts demanding cake and the parentals become completely oblivious. Let's make this a birthday Barnaby will *never* forget.”
created_at
"2026-02-20T19:43:46.503Z"
2/20/2026, 7:43:46 PM
model
"google/gemma-3-12b-it"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
223
moderation_latency
(null)
generation_time
13842
tokens_prompt
93
tokens_completion
808
native_tokens_prompt
75
native_tokens_completion
682
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
0.00009166
router
(null)
provider_responses
0
endpoint_id
"eb06dc92-5a16-47ec-a776-6ef956457c47"
id
"chatcmpl-RjmM0akql88h7Td17wsja59e"
is_byok
false
latency
58
model_permaslug
"google/gemma-3-12b-it"
provider_name
"DeepInfra"
status
200
api_type
"completions"
id
"gen-1771616626-imZdLewz8vhmjsEZ1Yi2"
upstream_id
"chatcmpl-RjmM0akql88h7Td17wsja59e"
total_cost
0.00009166
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
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: 68.32%, Deviation: 18.32%
neededClean
false
wordsTotal
483
wordsDialogue
330
21.6892%