NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYD7F21RC833Q0HPCERSVYBD
Run Details
Model:
MoonshotAI: Kimi K3 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
32.3s
Cost:
$0.011534
Provider:
Moonshot AI
Evaluation summary
Score
Evaluator
Details Link
0.08%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0424%
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
Molly spread the blueprints across the hood of Bill's rusted sedan, weighing down the corners with a brick, a donut, and Bill's hand. "Ow," said Bill. "Focus. The Henderson kid's seventh birthday. Two p.m. Saturday. Bounce house in the backyard, clown on retainer, and thirty-four confirmed presents." "How do you know it's thirty-four?" "I counted the RSVPs on the mother's social media. She posts everything, Bill. Everything. Last week she posted a photo of her lasagna with the security code visible on the alarm pad behind it." Bill squinted at the blueprints, which were actually crayon drawings Molly had commissioned from her nephew for two dollars and a juice box. "Why is the doghouse bigger than the house?" "Perspective isn't his strength. Listen, we go in as entertainers. You're the balloon animal guy." "I can't make balloon animals." "You've had three weeks to learn." Bill produced a mangled tube of latex from his pocket. "This was supposed to be a dog. The kids at the library cried." "It looks like a tax audit. Fine. New plan. You're the magician's assistant. I'm the magician." "Do you know any magic?" "I know how to make presents disappear." Molly wiggled her fingers. "That's the only trick we need. During the cake cutting, everyone's watching the kid blow out candles. Thirty seconds of darkness. We sweep the gift table into the bounce house, deflate it, and walk out with everything bundled like Santa's sack." "Where do we get a deflating device?" "It's called unplugging it, Bill." "Right. Right." He paused, peeling his hand out from under the donut. "Molly. Are we bad people? Stealing from a child?" "We're stealing from thirty-four children, technically. And no. We're redistributing. That kid already has a trampoline. I checked the registry. He has two trampolines." "Two?" "One's a backup. Rich people, Bill." They stood in silence, contemplating the crayon doghouse. "What about the clown?" Bill asked. "The real one." "Bubbles the Clown. I've studied his routine. He does forty-five minutes, takes a smoke break behind the garage, then does balloon swords. We time the grab during the smoke break." "What if he comes back early?" "Then you improvise." "I don't improvise. I panic and describe things. Last time I panicked I narrated a guard's haircut for six minutes." "It was a memorable haircut." "He had a gun, Molly." "A memorable haircut and a gun." She rolled up the blueprints and tucked them into her waistband like a general who'd lost every war but kept excellent maps. "Saturday, one-thirty. Wear something festive but forgettable." "Like what?" "You know how at parties there's always one guy nobody remembers inviting?" "Yeah." "Be that guy. That's your whole gift, Bill. You've been training for this your entire life." Bill nodded slowly, tucking the donut into his jacket for later. "You know, for once, I feel like the universe prepared me for something." "That's the spirit. Now get in the car. We need to case the party store." "For supplies?" "For a decoy piƱata. We're going to stuff it with our old parking tickets. Let the kid learn about consequences early." "That's actually kind of generous." "We're complicated people, Bill. Get in."
created_at
"2026-07-25T18:09:12.768Z"
7/25/2026, 6:09:12 PM
model
"moonshotai/kimi-k3-20260715"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2208
moderation_latency
(null)
generation_time
32252
tokens_prompt
75
tokens_completion
842
native_tokens_prompt
147
native_tokens_completion
766
native_tokens_completion_images
(null)
native_tokens_reasoning
16
native_tokens_cached
147
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
(null)
usage
0.0115341
router
(null)
provider_responses
0
endpoint_id
"650c9b12-7b6d-444d-8d92-df8daef84d7a"
id
"chatcmpl-6a64fbb857c78a5529a94a1b"
is_byok
false
latency
2208
model_permaslug
"moonshotai/kimi-k3-20260715"
provider_name
"Moonshot AI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1785002952-5XrxyRAY0JSdd2z9wZ6M"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1785002952-UTpR8CQkr0SG7LOTPi7K"
upstream_id
"chatcmpl-6a64fbb857c78a5529a94a1b"
total_cost
0.0115341
cache_discount
0.0003969
upstream_inference_cost
0
provider_name
"Moonshot AI"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
0.08%
Matches word count
n/a
neededClean
false
words
529
0.00%
Dialogue to Total Word Ratio
Ratio: 78.46%, Deviation: 48.46%
neededClean
false
wordsTotal
534
wordsDialogue
419
0.0424%