NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-02-25-01KJBHJDG9Y414NXN3WMXMP27X
Run Details
Model:
Qwen 3.5 27B
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
3m 5s
Cost:
$0.023867
Provider:
Alibaba
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
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
The suburban house glowed with neon balloons and the distinct hum of a thousand sugar-rushed toddlers. Molly adjusted her night vision goggles, squinting at the blueprint spread across the hood of their getaway van. Bill leaned against the bumper, chewing on a toothpick that was definitely not for smoking. The moon was hidden behind thick clouds, providing perfect cover for their illicit intentions. The van idled quietly, a black shark in a sea of manicured lawns. So the target is the living room? Bill asked, tapping the paper with a grease-stained finger. The living room is where the wrapping paper is. Molly corrected him, her voice low and steady. We are not here for the cake. We are here for the boxed electronics. Bill sighed, looking at the house. The party was in full swing. There were probably twenty kids running around screaming inside the walls. That was not a stealth operation. That was a war zone. It is a distraction. Molly pointed at the driveway. The parents were inside managing the chaos. The security system was set to holiday mode, which meant it ignored motion between the hours of three and five. Holiday mode? Bill raised an eyebrow. You mean to tell me the alarm system has a nap schedule? It is a smart home hub. Molly shrugged. It thinks everyone is asleep because the lights are dimmed for the cake cutting. We slip in during the candle lighting. Bill laughed, a dry, wheezing sound. You want to steal birthday presents while a child blows out candles? That is poetic. It is tactical. Molly rolled her eyes. We grab the boxes, leave the cake, and vanish before the first present is opened. What if someone finds out? Bill asked, his voice dropping. I mean, these are kids. If we take a toy, we are basically stealing their soul. We are not taking the toys. Molly snapped her fingers. We are taking the new gaming consoles and the tablets. The toys are just filler. Bill looked at the house again. The balloons were drifting down the street, caught in the wind. It felt wrong. He had never stolen from a birthday party before. He had stolen from banks, museums, even a casino, but this felt different. The stakes were low, but the moral weight was heavy. Do you think we should go? Bill asked, his voice trembling slightly. We already parked the van. Molly checked her watch. We have ten minutes before the cake comes out. Bill nodded slowly. He adjusted his gloves. Okay. Let us do this. But if I get hit by a plastic sword, I am suing you. Molly smiled, a sharp, dangerous thing. Deal. Now move. They crept toward the back door, the sound of laughter drifting through the open windows. The plan was solid, the timing was perfect, and the loot was wrapped in shiny paper. It was the perfect crime, if you ignored the fact that they were robbing a seven-year-old named Timmy. The door clicked open. Inside, the smell of frosting filled the air. Molly signaled Bill to move to the left. He nodded, checking his bag. It was going to be a long night, but at least the loot was colorful.
created_at
"2026-02-25T23:18:57.808Z"
2/25/2026, 11:18:57 PM
model
"qwen/qwen3.5-27b-20260224"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
617
moderation_latency
(null)
generation_time
185133
tokens_prompt
77
tokens_completion
10300
native_tokens_prompt
76
native_tokens_completion
9935
native_tokens_completion_images
(null)
native_tokens_reasoning
9256
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.0238668
router
(null)
provider_responses
0
endpoint_id
"f3f88554-0c22-4d57-973b-0dec948808e8"
id
"chatcmpl-01be37b5-d56d-9ba5-b39c-77749e17915d"
is_byok
false
latency
617
model_permaslug
"qwen/qwen3.5-27b-20260224"
provider_name
"Alibaba"
status
200
api_type
"completions"
id
"gen-1772061537-9WNNRwI9Lp6U6DAJss5S"
upstream_id
"chatcmpl-01be37b5-d56d-9ba5-b39c-77749e17915d"
total_cost
0.0238668
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Alibaba"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
539
0.00%
Dialogue to Total Word Ratio
Ratio: 0.00%, Deviation: 30.00%
neededClean
false
wordsTotal
543
wordsDialogue
0
0.0000%