NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-24-01M39WV6ZBE0V61Q8WP2KD35ZD
Run Details
Model:
Aion 3.5 Mini (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
3m 39s
Cost:
$0.014978
Provider:
AionLabs
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
64.54%
Dialogue to Total Word Ratio
(details)
32.2715%
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 had cased banks, museums, and one deeply regrettable aquarium, but she had never surveilled a bouncy castle before. This one was shaped like a dragon, deflating slowly at the left nostril, wheezing like an old man on stairs. She'd stolen a Fabergé egg once. Somehow this felt harder. The egg hadn't come with parents. She passed the binoculars to Bill, who slumped in the passenger seat wearing a ski mask in July because Molly insisted on professionalism. Through the lenses, the backyard resolved into chaos: eleven children, two folding tables, and a clown making balloon animals with the serene efficiency of a surgeon. "Security's tight," Molly said. "The security is a woman in a birthday hat holding a juice box." "The clown is the alarm system. You get near the present table, he honks. Kids scatter, parents look up, we're made." Bill lowered the binoculars and studied the clown. The van at the curb promised Certified Fun. Bill distrusted anyone who certified themselves in anything, but he had to admit the balloon giraffe was flawless. Then there was the piñata: a donkey hanging from the oak, turning slowly in the breeze like a warning to sailors. Molly had briefed him on piñata protocol for two hours the night before. It was, she explained, the guard dog of the party — blind, leashed, and packed with enough sugar to trigger a stampede if mishandled. Bill rubbed his eyes under the mask. "Walk me through it again. We have forty years of professional thieving between us, and the job is stealing birthday presents. From children." "From one child. The client wants the whole haul." "Who's the client, Mol?" Molly hesitated, which was rare. "Tiffany. Age six. The birthday girl's preschool rival." "We're doing a job for a kindergartner." "She pays in a jar of coins and three gummy worms. Best rate we've had all year." The gummy worms had melted into a single multicolored blob. But the coins were real, and business was business, even when the business was morally bankrupt. Molly laid the blueprint across the dashboard — drawn in crayon on a Chuck E. Cheese placemat, which she claimed was for authenticity. Phase one: Bill contained the piñata. Phase two: Molly neutralized the clown. Phase three: garbage bags, presents, exit. "You're on piñata duty," she said. "Piñata duty." "Keep it calm. Talk to it if you have to." Bill wanted to point out that the museum guard in Lisbon had been a person and this was a papier-mâché donkey, but arguing with Molly was like arguing with weather. He took the binoculars instead — and froze. A small girl in a party hat was staring directly at the van. Unblinking. Patient. She had the cold stillness of a professional and had been there for at least two minutes. "Mol." "I see her." Kids that age had no peripheral vision and no fear. This one would either rat them out or ask to join the crew. The girl raised one fat little hand and waved. Slowly. Deliberately. Like a mob boss acknowledging a debtor. "Is she recruiting us?" Molly started the engine. "We're leaving." The van peeled away in a screech of regret. Behind them, the dragon castle wheezed out its final breath. Bill pulled off his mask. "We're quitting a six-year-old's heist because another six-year-old outplayed us?" "We're professionals, Bill. Not philosophers." Molly checked the mirror. The girl was still watching, party hat perfectly level. "And we never work two sides of a birthday party."
created_at
"2026-09-24T14:24:22.524Z"
9/24/2026, 2:24:22 PM
model
"aion-labs/aion-3.5-mini-20260923"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1411
moderation_latency
(null)
generation_time
218376
tokens_prompt
75
tokens_completion
10120
native_tokens_prompt
65
native_tokens_completion
10690
native_tokens_completion_images
(null)
native_tokens_reasoning
9238
native_tokens_cached
65
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
0.0149777
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1790259862-7xpXqrtVdqQpgDDs1wDt"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790259862-e1Cn3ORAIW2S73HQyfrO"
upstream_id
"7197a22bf0154b0dab6b237bbfc2004c"
provider_responses
0
endpoint_id
"60ac3bdc-90ec-4769-9193-2c00b4d79562"
id
"7197a22bf0154b0dab6b237bbfc2004c"
is_byok
false
latency
1410
model_permaslug
"aion-labs/aion-3.5-mini-20260923"
provider_name
"AionLabs"
status
200
total_cost
0.0149777
cache_discount
0.0000338
upstream_inference_cost
0
provider_name
"AionLabs"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
583
64.54%
Dialogue to Total Word Ratio
Ratio: 27.04%, Deviation: 2.96%
neededClean
false
wordsTotal
588
wordsDialogue
159
32.2715%