NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-24-01M39X8VJAE5AXGAMGTHJ44GRJ
Run Details
Model:
Aion 3.5 Mini (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
2m 51s
Cost:
$0.013510
Provider:
AionLabs
Evaluation summary
Score
Evaluator
Details Link
86.38%
Matches word count
(details)
94.19%
Dialogue to Total Word Ratio
(details)
90.2871%
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 peered through the windshield of the stolen ice cream truck, binoculars pressed so hard against her eyes she could see the future—or at least the bouncy castle. Across the asphalt, the Sunnyvale Community Center was decorated to the nines. Streamers hung like festive vines, and a balloon arch stood guard over the entrance, looking less like a security measure and more like a colorful, inflatable threat. To the untrained eye, it was a birthday party. To Molly, it was a fortress waiting to be cracked. The sun beat down on the parking lot, making the colorful scene look even more deceptive. “Target acquired,” she whispered. “The Sterling kid’s seventh birthday. Rich parents, massive gift pile, one bored teenager in a wizard hat.” Bill, crammed into the passenger seat next to a stack of empty waffle cones, looked up from his sandwich. He had the look of a man who had seen too much and eaten too much grease. His turtleneck, usually his lucky charm, felt suddenly very tight. “We’re really doing this? We’re stealing from a kid?” “It’s a legacy. That dragon is worth more than my car.” Bill took a bite, chewing thoughtfully. “It’s a turkey club. Want some?” “You are disgusting.” “I’m carbo-loading. Heists require energy.” Molly ignored him and unrolled the blueprints on the dashboard. The layout was simple: a large room, a bouncy castle in the corner, and a mountain of presents stacked haphazardly near the punch bowl. The whole thing looked like a trap designed by someone who hated both architecture and children. She traced a path with a red pen, marking the blind spots behind the gift table. “The defense grid is simple. A balloon arch, a bouncy castle, and a teenager named Chad.” “Chad?” “He’s the bouncer. Literally. He bounces. If you cross him, you’re going into the ball pit.” Bill shuddered. He had a complicated history with ball pits, involving a lost shoe and a mild existential crisis. “What about the clown?” “Giggles is a wildcard. We avoid eye contact. If he approaches, you throw the turkey club at him and run.” “And the kids?” “They’re hopped up on sugar. They’ll think we’re the ‘Presents Fairies’.” Bill pulled his ski mask on. He looked like a sad, lumpy egg. “Fine. But if anyone cries, I’m out.” “No crying. We grab the loot and vanish. Twenty minutes, tops.” Bill looked at the balloon arch again. It swayed gently in the breeze, a sentinel of doom. He thought about his retirement plan, which mostly involved winning the lottery or faking his own death. This job paid well, but the hazard pay for potential glitter bombs was never enough. Molly checked her watch. The magician was due to start his act in five minutes, which would provide the perfect distraction. Bill sighed, a sound of pure defeat mixed with professional resignation. “Let’s go steal some toys.” They stepped out of the van and into the bright sunlight, two master criminals walking toward a party, ready to commit the most ridiculous crime of their careers.
created_at
"2026-09-24T14:31:49.59Z"
9/24/2026, 2:31:49 PM
model
"aion-labs/aion-3.5-mini-20260923"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
822
moderation_latency
(null)
generation_time
171262
tokens_prompt
75
tokens_completion
10247
native_tokens_prompt
65
native_tokens_completion
9642
native_tokens_completion_images
(null)
native_tokens_reasoning
9447
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
0.0135105
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1790260309-Z6TCR8JVHYQeQWgTePTL"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790260309-3fCq6K7NyRpiTMAjQ2nr"
upstream_id
"7fa8e82ab5a24dc09e6884acbc50196f"
provider_responses
0
endpoint_id
"60ac3bdc-90ec-4769-9193-2c00b4d79562"
id
"7fa8e82ab5a24dc09e6884acbc50196f"
is_byok
false
latency
822
model_permaslug
"aion-labs/aion-3.5-mini-20260923"
provider_name
"AionLabs"
status
200
total_cost
0.0135105
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
86.38%
Matches word count
n/a
neededClean
false
words
511
94.19%
Dialogue to Total Word Ratio
Ratio: 28.91%, Deviation: 1.09%
neededClean
false
wordsTotal
512
wordsDialogue
148
90.2871%