NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHY96WYADX5B1BSHCK70FE0R
Run Details
Model:
Gemma 3 4B
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
7.6s
Cost:
$0.000051
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
“Seriously, Bill? A children’s birthday party?” “Don’t look so horrified, Molly. It’s… lucrative.” “Lucrative? You’re describing a room full of five-year-olds who haven't yet learned the concept of ‘personal property.’ This is a nightmare scenario.” “Think of it! Mountains of brightly wrapped packages. Perfect distribution of value. We’re talking bespoke LEGO sets, remote-control dinosaurs, sparkly unicorn headbands… This isn't ‘nightmare’ this is ‘dream’." “You're awfully enthusiastic about glitter. And miniature plastic riding ponies.” “Okay, okay, let’s talk logistics. The caterer’s is right by the bouncy castle, which is directly adjacent to the main party area. Windows are small, single-paned. We’ll need a shim.” “A shim? You're suggesting we use a shim to breach a window teeming with sugar-fueled toddlers? That sounds remarkably… inefficient.” “It’s a *small* shim, Molly. Think of it as a… diplomatic solution. Plus, the parents will be distracted by the face painting. Guaranteed.” “Face painting? Seriously? Little Timmy is going to have a butterfly on his face while we’re liberating a truckload of stuffed animals.” “It’s crucial intel, Molly! Butterfly face painting equals maximum parental distraction. Trust me.” “I don't ‘trust’ you, Bill. I tolerate you. And the tolerance level is currently plummeting.” “Look, the spreadsheet says we’re looking at roughly seventy-two presents. Minimum. We could fund our retirement for a decade with that.” “Retirement? We’re twenty-eight, Bill! We're supposed to be stressing about mortgages, not pirate costumes.” “Pirate costumes are valuable, Molly. Remember that Captain Hook outfit we found at the estate sale?” “That was a *bargain*, Bill. Not a target. This is about presents, pure and simple. And the sheer, perplexing joy these children get from them.” “Joy is overrated. Now, about the diversion... I was thinking a strategically placed flock of pigeons?” “Pigeons? You want to unleash a swarm of bewildered pigeons into a room full of brand-new, expensive toys? Bill, that’s… spectacularly bad.” "Okay, okay, scratch the pigeons. How about a rogue mariachi band?" “A mariachi band? At a children’s party? You're actively trying to case our plans as a disaster, aren't you?” “Just considering options, Molly! Think of the chaos! The surprise! The... proceeds!” “Let's stick to the shim, Bill. And please, for the love of all that is sensible, no pigeons.”
created_at
"2026-02-20T19:42:44.18Z"
2/20/2026, 7:42:44 PM
model
"google/gemma-3-4b-it"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
63
moderation_latency
(null)
generation_time
7468
tokens_prompt
93
tokens_completion
694
native_tokens_prompt
75
native_tokens_completion
602
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
0.00005116
router
(null)
provider_responses
0
endpoint_id
"d3de3bd4-81bc-48fb-924f-2a87b2a36e75"
id
"chatcmpl-RTkq8lrmjtE87nKpDW752A4w"
is_byok
false
latency
8
model_permaslug
"google/gemma-3-4b-it"
provider_name
"DeepInfra"
status
200
api_type
"completions"
id
"gen-1771616564-xJtXt6N2nJPWHrFJ2Qyg"
upstream_id
"chatcmpl-RTkq8lrmjtE87nKpDW752A4w"
total_cost
0.00005116
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
378
0.00%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 30.00%
neededClean
false
wordsTotal
378
wordsDialogue
378
0.0000%