NC Bench
Overview
About
Tests
Models
Model
Gemini 3 Flash (Preview)
Test
Dialogue tags
Scenario
Write 500 words with 70% dialogue
Run ID
2026-02-08-01KGYXK86WPPK7ZFM3ZFW7YYAB
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
"Check your six, Bill. We’ve got a bogey at three o’clock." "That’s a golden retriever, Molly. He’s sniffing a hydrangea." "He’s a scout, Bill. They use them for perimeter defense. Now, stay low. If the bouncy castle shifts even an inch, we lose our cover." "I am currently face-down in a bed of mulch that smells like fermented hamster bedding. How much lower do you want me to go?" "Hush. We are here for the loot. Look at that haul. The mountain of primary colors. The crinkle of high-end wrapping paper. It’s glorious." "It’s a five-year-old’s birthday, Molly. We are literally stealing Lego sets and probably a few rogue jars of slime." "Lego has a fantastic resale value, Bill! Do you know what a mint-condition Star Wars Ultimate Collector Series Millenium Falcon goes for on the black market?" "There is no way a kid named Tyler has an UCS Falcon at his 'Spiderman and Friends' garden party." "Don’t underestimate Tyler. His dad is a hedge fund manager. That stack of boxes is worth more than our last three jewelry heists combined, and it’s guarded by nothing but a clown named Sparkles who looks like he’s having a mid-life crisis." "Sparkles has a balloon pump, Molly. Those things can be pressurized." "Are you afraid of a man in size twenty-four shoes?" "I’m afraid of the legal ramifications of being tackled by a man in size twenty-four shoes. The headlines alone would kill me." "Focus. The cake is coming out. This is the diversion we discussed. When they start the off-key rendition of 'Happy Birthday,' we move. I take the north flank behind the sandbox. You crawl through the tunnel of the play-fort." "The tunnel is plastic, Molly. I’m six-foot-two. I’ll get wedged in there like a cork in a bottle." "Then use the slide as a tactical ramp! Honestly, Bill, where is your sense of adventure?" "It’s buried under the realization that I’m wearing a tactical turtleneck to steal a Paw Patrol Lookout Tower." "Shut up, they’re lighting the candles. Look at the distraction. The mother is crying, the father is trying to find a lighter, and Sparkles is making a poodle out of a 260Q balloon. Go, go, go!" "My knees are clicking! Does everyone hear that? It sounds like a bag of marbles!" "It’s drowned out by the screaming, Bill! Children are nature’s white noise. Grab the big box with the silver bow. That’s the high-value target." "I’ve got it. It’s heavy. Is this a rock? Did someone give this kid a boulder?" "It’s probably a professional-grade telescope. Or a solid gold tricycle. Don't drop it!" "Molly, wait. There's a label. It says 'To Tyler, love Grandma.'" "A grandmother’s love is priceless, Bill. Now get to the getaway van before the toddler juice-boxes kick in and they become a mindless horde." "If we get caught, I’m telling the cops you’re my therapist and this is an extreme form of exposure therapy." "Just run, you idiot! The clown is looking right at us!"
created_at
"2026-02-08T15:23:26.034646+00:00"
Sunday, February 8th, 2026 at 3:23:26 PM GMT+00:00
model
"google/gemini-3-flash-preview-20251217"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
679
moderation_latency
(null)
generation_time
6799
tokens_prompt
61
tokens_completion
684
native_tokens_prompt
67
native_tokens_completion
748
native_tokens_completion_images
0
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
0.0022775
router
(null)
provider_responses
0
status
200
is_byok
false
latency
679
endpoint_id
"6e82f652-2693-4b60-bbe0-7362d6e345f0"
provider_name
"Google AI Studio"
model_permaslug
"google/gemini-3-flash-preview-20251217"
api_type
"completions"
id
"gen-1770564198-xnEZji8gQFTS0WO5Xlgj"
upstream_id
(null)
total_cost
0.0022775
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Google AI Studio"
Evaluation details
Result
Evaluator
Details
Meta Data
35.0022%
Matches word count
n/a
neededClean
false
words
518
0.0000%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 30.00%
neededClean
false
wordsTotal
518
wordsDialogue
518
17.5011%