NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCTKQCN8HFWN5KNDHKQ09WX
Run Details
Model:
Gemini 3.6 Flash (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
1m 4s
Cost:
$0.119298
Provider:
Google AI Studio
Evaluation summary
Score
Evaluator
Details Link
95.99%
Matches word count
(details)
99.99%
Dialogue to Total Word Ratio
(details)
97.9895%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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 damp basement smelled of burnt espresso, cheap plastic, and unbridled ambition. Molly unrolled a blueprint across the rickety folding table, weighing down the curled edges with a half-eaten bagel and a box of apple juice. She leaned forward, her tactical flashlight illuminating the map. Bill squinted through the dim light at the complex schematic, which was noticeably drawn in pastel pink crayon and covered in a thick dusting of gold glitter. Bill pointed a heavy finger at a chaotic zigzag across the back porch. "Is that a lethal security grid?" Molly leaned in close, her right eye twitching with focus. "Worse. Party streamers. Silent, but devastating to momentum." Bill rubbed his face, wondering where his life had gone wrong. "Molly, it’s a seven-year-old’s birthday party." "It’s an inflated fortress of high-value cargo! Do not underestimate the enemy." She began pacing around the swinging lightbulb, tapping a wooden spoon against her palm like a seasoned general preparing for war. Outside, the quiet suburban cul-de-sac slept, blissfully unaware of the tactical strike brewing beneath them. Molly dramatically pulled a heavy velvet cloth off the center table, revealing their gear: two neon rubber sticky hands, a homemade baking soda smoke bomb, a pair of night-vision goggles bought off the internet, and a stuffed alligator. Bill crossed his arms and looked at the strange pile. "What’s the plush reptile for?" Molly gestured wildly over the map. "Distraction. When the clown enters Sector B, you drop the gator into the punch bowl. Parents scramble. We breach the table." "They have an operator named Greg. He has a whistle." Molly slid a crisp brown envelope across the table with a smirk. "I neutralized Greg. A fake parking ticket will keep him occupied for twenty minutes." Bill sighed deeply. He had cracked sophisticated steel safes in Berlin, evaded bloodhounds in Munich, and bypassed laser corridors in Antwerp, yet here he was, sweating over a tactical approach to a pin-the-tail-on-the-donkey game. The stakes felt absurdly high, mostly because Molly had guaranteed him fifty percent of whatever high-end Lego sets they successfully extracted from the premises. He adjusted his dark turtleneck and stared at the crayon map with growing anxiety. "What about the primary target?" Bill asked, leaning over the blueprint. "The mothership. Wrapped in shiny paper. Intelligence says it’s a motorized mini Jeep." "We’ll need a bigger getaway vehicle." "Already handled. I bribed the paperboy for his red wagon." "If we get caught, my reputation is ruined." "If we pull this off, we own the sandbox." Molly checked her wristwatch, synchronizing it with the ticking clock on the wood-paneled wall. The hour of the invasion was rapidly approaching. She tossed a black knit balaclava toward Bill, who caught it with a long, defeated groan. Embroidered directly onto the dark fabric forehead was a tiny, brightly colored glittery butterfly. "Is the glittery butterfly necessary for stealth?" Bill asked, inspecting the patch. "It instills terror. Put it on and move out."
created_at
"2026-07-25T14:24:34.21Z"
7/25/2026, 2:24:34 PM
model
"google/gemini-3.6-flash-20260721"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1679
moderation_latency
(null)
generation_time
64166
tokens_prompt
75
tokens_completion
4483
native_tokens_prompt
67
native_tokens_completion
15893
native_tokens_completion_images
0
native_tokens_reasoning
15212
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
0.119298
router
(null)
provider_responses
0
endpoint_id
"5d6d133d-a953-4781-a835-7ee76f4e1388"
id
"Isdkar_fE53i_uMP4bTtoAc"
is_byok
false
latency
1679
model_permaslug
"google/gemini-3.6-flash-20260721"
provider_name
"Google AI Studio"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1784989474-NWqd4YSA7cGet6QpooBJ"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784989474-B32NbDj4DjosEHzHqBRa"
upstream_id
"Isdkar_fE53i_uMP4bTtoAc"
total_cost
0.119298
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Google AI Studio"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
95.99%
Matches word count
n/a
neededClean
false
words
492
99.99%
Dialogue to Total Word Ratio
Ratio: 30.04%, Deviation: 0.04%
neededClean
false
wordsTotal
506
wordsDialogue
152
97.9895%