NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-11-19-01M7SVXBS4761DJ3XCZEE6D1B7
Run Details
Model:
Gemini 4 Hydrogen (Reasoning, Medium)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
49.8s
Cost:
$0.056928
Provider:
Google
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
93.60%
Dialogue to Total Word Ratio
(details)
96.7597%
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
Bill slapped a laminated map onto the sticky diner table, narrowly missing Molly's plate of lukewarm curly fries. "Alright, listen up," Bill said, tapping a red circle with his plastic fork. "We hit the Henderson residence at fourteen hundred hours. Maximum capacity will be twenty-five six-year-olds, twelve exhausted parents, and one freelance magician." Molly chewed a fry slowly, squinting at the neon-colored paper beneath his fork. "Is this a real blueprint, or did you steal a placemat off their kitchen table?" "It is a tactical schematic," Bill snapped, smoothing down a crease. "Six-foot cedar fence, one side gate, guarded by a golden retriever named Barnaby. According to my intel, Barnaby is easily bribed with thick-cut bacon." "We are professional jewel thieves, Bill," Molly said, rubbing her temples. "Why are we robbing a first-grader named Timmy?" "Because Timmy's dad owns three car dealerships, and the invite said no boxed gifts, which means envelopes of cash," Bill whispered, leaning across the vinyl booth. "Plus, word on the street is Timmy's grandma bought him a mint-condition vintage train set." "A train set," Molly repeated flatly. She pushed her plate away and stared at her partner. "I cracked the vault at the Antwerp Diamond Exchange." "And you got caught because you sneezed into a laser grid," Bill countered, leaning back. "We need a soft score. Phase one is infiltration. I have disguises." "Please tell me we are posing as the catering staff." "Better. We are the entertainment. You are Bubbles the Clown, and I am your mute assistant, Sad Trombone." Molly stared at him for a long, heavy moment. "I am not wearing oversized shoes, Bill. My arches need support." "You will wear the shoes, Molly, because nobody suspects a clown holding a giant burlap sack," Bill insisted. "While I make substandard balloon animals, you slip into the sunroom and drop the loot down the laundry chute." "What about the freelance magician?" "I looked him up," Bill said, pulling a crumpled flyer from his jacket pocket. "The Great Zucchini. Does basic card tricks. If he interferes, I will neutralize him." "Please do not assault a birthday magician." "I meant verbally, Molly. I will heckle him until he cries and leaves. It is a classic misdirection technique." Molly sighed, reaching back for one last fry. "And what happens when twenty-five sugar-crazed children realize the presents are gone? Toddlers in a blind rage have no moral compass, Bill. They bite." "That is why we execute the extraction during the piƱata ceremony," Bill said proudly, folding his hands. "When that donkey bursts, it will be pure chaos. We walk right out the front door." "You have actually thought this through," Molly admitted, a reluctant smile tugging at the corner of her mouth. She picked up her coffee mug and clinked it against his water glass. "One condition. I get to keep the train set." "Deal," Bill said, sliding out of the booth and grabbing the check. "Now finish your fries. We have to go practice tying poodles out of latex."
created_at
"2026-11-19T12:18:42.602Z"
11/19/2026, 12:18:42 PM
model
"google/gemini-4-hydrogen-20261112"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1273
moderation_latency
(null)
generation_time
49806
tokens_prompt
75
tokens_completion
2715
native_tokens_prompt
97
native_tokens_completion
7571
native_tokens_completion_images
0
native_tokens_reasoning
6870
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
0.056928
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.22.2; linux; x64))"
http_referer
(null)
request_id
"req-1795090722-Fj27vmEV13M1A2r4QrZw"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1795090722-2Umm0O5cwjbIndln5DDp"
upstream_id
"Iun-atXqJ4WbotQPpeSEsA8"
provider_responses
0
endpoint_id
"3936f89d-7d96-43b5-adaf-f690b9de1298"
id
"Iun-atXqJ4WbotQPpeSEsA8"
is_byok
false
latency
1273
model_permaslug
"google/gemini-4-hydrogen-20261112"
provider_name
"Google"
status
200
total_cost
0.056928
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
99.92%
Matches word count
n/a
neededClean
false
words
503
93.60%
Dialogue to Total Word Ratio
Ratio: 71.15%, Deviation: 1.15%
neededClean
false
wordsTotal
513
wordsDialogue
365
96.7597%