NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHY804VBJGNBZY0N9AACJ0N3
Run Details
Model:
Gemma 3 12B
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
23.7s
Cost:
$0.000097
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
9.61%
Matches word count
(details)
36.71%
Dialogue to Total Word Ratio
(details)
23.1582%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 aroma of sugar and impending doom hung thick in Molly’s tiny apartment. Bill, a man whose enthusiasm for larceny was only rivaled by his love of novelty socks, was bouncing on the floral sofa. He was currently sporting ones with cartoon badgers playing croquet. “So, you’re thinking… a children’s party?” Bill asked, eyes wide. “Molly, are you sure this isn’t a *bit* low-rent for us?” Molly, meticulously applying a shade of purple lipstick she referred to as “Strategic Disguise,” didn't look up. “Low-rent keeps us under the radar, Bill. High-profile gallery heists require a certain… *flair*. This requires stealth and a strong tolerance for glitter.” “Glitter. Right. Because nothing says ‘professional criminal’ like having sparkly residue clinging to your eyebrows.” Bill dramatically brushed an invisible speck from his cheek. “Think about it. Do you seriously believe security at a five-year-old’s birthday party involves laser grids and infrared sensors?” Molly finally glanced at him, eyebrow arched. “We’re talking bouncy castles and a pinata shaped like a unicorn.” “A unicorn pinata, you say? Is it… sturdy?” Bill asked seriously. Molly sighed, fighting the urge to strangle him with a string of fairy lights. “Bill! Presents. We’re stealing presents. High-end, kid-friendly presents. Think robot dinosaurs, limited edition dolls, miniature ponies… Parents these days spend a fortune.” “Okay, okay, presents. Good point. What’s the theme? Pirate? Princess? Because I’ve got a surprisingly convincing pirate eye patch if we need it.” “It’s ‘Enchanted Forest,’” Molly stated, rolling her eyes. "Which means an awful lot of woodland creatures and probably an unnerving amount of tulle." “Tulle. My nemesis. It clings. It snags. It mocks me with its sheer existence.” Bill shuddered. “So, the plan? Do we go in as parents? I could do a convincing dad who’s slightly overwhelmed and desperately needs a coffee.” “Even you would be a bit *too* convincing, Bill. No. We’re going in as performers.” Molly finished her lipstick. “I've got a contact— a guy who runs a children’s entertainment company. He's getting us booked to do a ‘magical forest’ performance.” “You’re getting us booked… as entertainers? Molly, I can barely juggle three beanbags, let alone conjure a rabbit out of a hat!” “You don't have to conjure anything. You’re a ‘jester’. You wear a silly hat and tell bad jokes. I’ll be the ‘enchantress’. I’ll distract with sparkles and disappearances.” Molly grinned. "And when the parents are mesmerized by my illusions, you swoop in and liberate the loot." Bill considered this, stroking his chin. “And the worst part? They’ll probably *tip* us. A five-year-old handing over a crumpled dollar bill… it’s just… beautiful, isn't it?" “Absolutely beautiful,” Molly agreed, grabbing a sequined cape. “Now, find me a jester hat with extra bells. And for the love of all that is holy, please try not to trip over the unicorn pinata."
created_at
"2026-02-20T19:21:34.327Z"
2/20/2026, 7:21:34 PM
model
"google/gemma-3-12b-it"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
642
moderation_latency
(null)
generation_time
23585
tokens_prompt
93
tokens_completion
819
native_tokens_prompt
75
native_tokens_completion
721
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
0.00009673
router
(null)
provider_responses
0
endpoint_id
"eb06dc92-5a16-47ec-a776-6ef956457c47"
id
"chatcmpl-RBhGhUH5F3rg0NLyraQP1XxQ"
is_byok
false
latency
416
model_permaslug
"google/gemma-3-12b-it"
provider_name
"DeepInfra"
status
200
api_type
"completions"
id
"gen-1771615294-aurhmIk4H7eDzfXLKq9e"
upstream_id
"chatcmpl-RBhGhUH5F3rg0NLyraQP1XxQ"
total_cost
0.00009673
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
9.61%
Matches word count
n/a
neededClean
false
words
478
36.71%
Dialogue to Total Word Ratio
Ratio: 74.48%, Deviation: 4.48%
neededClean
false
wordsTotal
478
wordsDialogue
356
23.1582%