NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-11-19-01M7SQ8BAY0MX253D7VNTFE12Y
Run Details
Model:
Gemini 4 Neon (Reasoning, Medium)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
58.9s
Cost:
$0.105231
Provider:
Google
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
98.27%
Dialogue to Total Word Ratio
(details)
99.1357%
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
Molly unrolled the crinkled parchment across the kitchen table, weighing down the corners with three half-empty cans of cheap soda and a stale bagel. The diagram showed a sprawling suburban backyard, meticulously labeled in red marker. Bill leaned over the map, squinting through the haze of a flickering fluorescent bulb. He tapped a large purple square near the patio. "Tell me again why we are hitting a seven-year-old's birthday instead of a jewelry store," Bill said, scratching his chin. "Because jewelry stores have laser grids and armed guards, Bill. Little Barnaby Henderson has a hired magician with a bad back and a mom who drinks too much pinot grigio," Molly replied, tapping the red X in the center of the living room. "The loot is sitting right there on the dining table. We are talking unopened vintage action figures, a top-tier gaming console, and at least four hundred dollars in crisp twenties tucked inside musical greeting cards." Bill picked up a plastic army man Molly was using to represent the target's golden retriever. He frowned at the tiny green figurine, turning it over in his calloused palms. He had spent six years cracking safes in downtown Chicago, and now he was calculating the patrol route of a mildly obese dog named Buster. "What about the perimeter? The intel says they rented an inflatable castle," Bill asked, setting the plastic soldier down near the patio door. "Those things create massive blind spots, and the kids inside are basically feral." "That is where you come in," Molly said, flashing a grin that usually preceded a terrible felony. "You are going in disguised as a balloon artist. You distract the pint-sized security force by twisting dachshunds while I slip past the snack table. Once the cake comes out, everyone migrates to the kitchen for the singing. That gives me a ninety-second window to bag the presents and extract through the laundry room window." Bill stared at her in absolute silence. Outside the apartment window, a siren wailed in the distance, reminding him of simpler times when they only stole luxury sedans. He reached for the stale bagel, broke off a piece, and chewed thoughtfully while picturing himself wearing suspenders and a bright red nose. "I do not know how to make a balloon dog, Molly. Last time I tried, I made a deformed worm and a toddler cried," Bill protested. "Then make worms and call them snakes. Kids love reptiles," Molly snapped, rolling up the blueprint with brisk efficiency. "Just keep the sugar-crazed mob occupied until you hear me honk the van twice. Do not eat the cupcakes, do not get trapped in the bounce house, and whatever you do, stay away from the face-painting station." Bill sighed heavily, already mourning the death of his criminal dignity. He stood, grabbed a crumpled yellow vest from the back of his chair, and shoved a handful of uninflated balloons into his pocket. "Fine. But if we pull this off, I am keeping the gaming console."
created_at
"2026-11-19T10:57:19.718Z"
11/19/2026, 10:57:19 AM
model
"google/gemini-4-neon-20261118"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1861
moderation_latency
(null)
generation_time
58898
tokens_prompt
75
tokens_completion
2050
native_tokens_prompt
97
native_tokens_completion
6996
native_tokens_completion_images
0
native_tokens_reasoning
6369
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"STOP"
service_tier
"default"
usage
0.105231
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.22.2; linux; x64))"
http_referer
(null)
request_id
"req-1795085839-nOujlwLks3zf4eLj3yLM"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1795085839-0gs1w9Np8Yrpbvcj2jsF"
upstream_id
"D9b-atCDMOOa4_UP9caAyA8"
provider_responses
0
endpoint_id
"cbc1d01f-e431-41ec-9abb-36c1a5b10cb4"
id
"D9b-atCDMOOa4_UP9caAyA8"
is_byok
false
latency
1860
model_permaslug
"google/gemini-4-neon-20261118"
provider_name
"Google"
status
200
total_cost
0.105231
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Google"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
500
98.27%
Dialogue to Total Word Ratio
Ratio: 50.59%, Deviation: 0.59%
neededClean
false
wordsTotal
508
wordsDialogue
257
99.1357%