NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-27-01M5YFHTGRB5XS53JJ2TD9TWF2
Run Details
Model:
Kimi K3.1 Pro (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
4m 4s
Cost:
$0.083082
Provider:
Moonshot AI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
92.19%
Dialogue to Total Word Ratio
(details)
96.0973%
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 unrolled the blueprint across Bill’s kitchen table, pinning one corner with a mug and the other with a suspiciously sticky cinnamon bun. Their previous job had involved an unattended charity shop and had still ended with Bill apologising to a hat stand. Molly had chosen the party because it offered a narrow window, minimal security, and enough wrapping paper to disguise a getaway. Also, the invitation mentioned a chocolate fountain, which she considered actionable intelligence. “This,” she said, “is not a blueprint. It’s a drawing of a castle with a dragon on it.” “The dragon is important,” Bill replied. “That’s where we enter.” “You enter a bouncy castle through the dragon?” “Through the mouth. Obviously.” Bill leaned back, looking wounded by the question. He had spent twelve years stealing things and still believed preparation meant owning a pencil. Molly tapped the drawing. Saturday’s party occupied a suburban garden currently defended by a low fence, several balloons, and a woman named Deirdre who had once frightened off a Jehovah’s Witness using only a casserole dish. “Fine. What about the clown, the magician, and thirty sugar-powered children?” Outside, rain needled the windows. Somewhere below, a car alarm hiccuped, stopped, then seemed to think better of it. On the table lay their equipment: two laundry sacks, a tube of glue, and Bill’s false moustache, which had been false in several previous identities and now looked tired of the whole profession. Molly crossed out ‘grappling hook’ and wrote ‘party bags’. Bill objected that party bags lacked drama. She reminded him that so did prison, except for the drama involving other people’s toilets. Neither mentioned the walkie-talkie shaped like a banana. “We go at two. The clown will be in the kitchen, crying into his wig.” “And the magician?” “Terrified of rabbits. We release Gerald.” He nodded toward Molly’s handbag, from which two sock ears protruded. She had wondered about those. “Gerald is my handbag.” “He has the ears for it.” Molly made a note. It read: Replace Bill. Then she turned the castle sideways and located something resembling reality. “The presents are here, beside the cake. We bag them, cross the ball pit, and exit through the hedge.” “Why the hedge?” “Because you drew a dragon where the gate should be.” Bill accepted this with the dignity of a man whose criminal genius had been mistaken for stationery. He lifted the cinnamon bun, regretted every decision since Tuesday, and used the blueprint to clean his fingers. “Fair. What if someone sees us?” “You’re wearing your stripy jumper. They’ll assume you’re entertainment.” He looked down. The jumper had seemed cheerfully anonymous in the shop. Now it suggested a man who might, at any moment, produce a balloon poodle or weep. “And if they ask for a show?” “Juggle.” “I can’t juggle.” Molly folded the plan, tucked Gerald under her arm, and smiled with the bright patience of a woman already composing her alibi. “Then we’ll finally have a reason to run, you magnificent turnip.”
created_at
"2026-10-27T10:47:30.082Z"
10/27/2026, 10:47:30 AM
model
"moonshotai/kimi-k3.1-pro-20261023"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
5957
moderation_latency
(null)
generation_time
243394
tokens_prompt
75
tokens_completion
8648
native_tokens_prompt
146
native_tokens_completion
8279
native_tokens_completion_images
(null)
native_tokens_reasoning
7593
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
(null)
usage
0.083082
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.22.1; linux; x64))"
http_referer
(null)
request_id
"req-1793098050-UjhbTxeMBnb8TB4QBoST"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1793098050-7iFbqEpizPhajMdxDI9t"
upstream_id
"chatcmpl-6ae0814727d7263b7959fe9a"
provider_responses
0
endpoint_id
"a3fafaac-f097-4ef1-86b3-842854dad5c6"
id
"chatcmpl-6ae0814727d7263b7959fe9a"
is_byok
false
latency
5957
model_permaslug
"moonshotai/kimi-k3.1-pro-20261023"
provider_name
"Moonshot AI"
status
200
total_cost
0.083082
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Moonshot AI"
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
92.19%
Dialogue to Total Word Ratio
Ratio: 31.27%, Deviation: 1.27%
neededClean
false
wordsTotal
502
wordsDialogue
157
96.0973%