NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHYAAF7DTWG2BGK9BE4TXA2S
Run Details
Model:
Gemma 3 27B
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
24.1s
Cost:
$0.000121
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
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
“Okay, so let’s go over this again, Bill. For the last time.” Molly tapped a chipped mug against a greasy diner tabletop. “No improvising. Absolutely none.” Bill slurped his coffee, leaving a brown mustache. “Improvisation is my middle name, Mol.” “It’s Arthur. Your middle name is Arthur.” “Details, details. Point is, I’m a quick thinker. A magpie, drawn to shiny opportunities.” “The shiny opportunity is a mountain of Paw Patrol backpacks and a remote control car shaped like a dinosaur. It's not exactly the crown jewels.” “Don’t underestimate the resale value of a limited-edition Skye plushie. Those things are going for a fortune on eBay.” Bill winked. “Alright, alright. No improv. Walk me through it.” “Right. Six-year-old Timmy’s birthday. We go in disguised as… what did we decide?” “A very enthusiastic clown and a slightly disgruntled balloon animal artist.” “Disgruntled is key, Bill. You look like you haven’t slept in a week. Authentic artistry requires suffering.” Bill sighed dramatically. “My soul *is* suffering, Molly. The things I do for a cut of Timmy's loot.” “It’s not loot, it’s…gifts. And it’s for Reginald. Remember Reginald?” “The ferret? Yes, Reginald needs a new heated bed. A very expensive heated bed, apparently.” “He gets cold easily. Alright. We scope out the gift table. Distraction is paramount. You’re in charge of the distraction.” “Me? Why me? You're the master of disguise!” “Because you have a regrettable knack for accidentally setting off fire alarms. I need you to *not* set off the fire alarm.” “Hey! That was one time! And it was a faulty smoke detector!” “It was because you were trying to juggle flaming bowling pins.” “They were plastic! And it looked impressive!” “Focus, Bill. The distraction. You’re going to ‘accidentally’ release all the balloons. Big, colourful chaos. Everyone looks up. I grab the gifts. We’re out.” “Release all the balloons? That feels… dramatic.” “It is dramatic. It's also effective. And it'll give the little tykes something to talk about besides my obviously fake nose.” “What if they catch us?” “They won’t. I’ve practiced my ‘helpful party entertainer’ face. It’s award-winning. You just look miserable and hold a lot of twisted balloon dogs.” “Miserable is easy. What if the parents are suspicious?” “Play dumb. ‘Oh, these delightful children! So much energy!’ and then mutter something about wanting a nap. People will assume you’re just a very tired artist.” “And the dog? Timmy’s got a golden retriever. He looks…alert.” “We’re not stealing from the dog, Bill. We’re stealing from the *children*. Though, if he tries to bite you, run.” “Right. Balloons, misery, no fire alarms. And Reginald gets a heated bed.” Bill finished his coffee. "This is a terrible plan." Molly grinned. “That’s why it’ll work.”
created_at
"2026-02-20T20:02:09.783Z"
2/20/2026, 8:02:09 PM
model
"google/gemma-3-27b-it"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
765
moderation_latency
(null)
generation_time
23968
tokens_prompt
93
tokens_completion
823
native_tokens_prompt
75
native_tokens_completion
719
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
0.00012104
router
(null)
provider_responses
0
endpoint_id
"8f22002c-c045-446f-a1b9-9896133536b8"
id
"chatcmpl-RCETWdvhhLe0efOsPfGj9HFi"
is_byok
false
latency
361
model_permaslug
"google/gemma-3-27b-it"
provider_name
"DeepInfra"
status
200
api_type
"completions"
id
"gen-1771617729-Uo8mlvQYGhD0JBGXJf0T"
upstream_id
"chatcmpl-RCETWdvhhLe0efOsPfGj9HFi"
total_cost
0.00012104
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
455
0.00%
Dialogue to Total Word Ratio
Ratio: 93.63%, Deviation: 23.63%
neededClean
false
wordsTotal
455
wordsDialogue
426
0.0000%