NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KY6YS6DSJ8JSKQTZQGF7CS
Run Details
Model:
Hy4 Preview (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
5m 18s
Cost:
$0.059110
Provider:
Tencent
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
100.00%
Dialogue to Total Word Ratio
(details)
100.0000%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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 getaway minivan crouched at the curb like a guilty turtle. Across the street, the Pinchbottom house glowed with paper lanterns, a bouncing castle, and a banner that read HAPPY BIRTHDAY, TRAKTOR, in letters slightly too large for a six-year-old. Molly studied the scene through binoculars while Bill tried to inflate a whoopee cushion without passing out. Molly’s blueprint of the house was spread across the dashboard, annotated in three colors of ink and one crayon. A half-eaten granola bar had fused itself to the passenger seat, a silent warning to anyone who considered sitting there. Molly lowered the binoculars. “Target acquired. Traktor’s party has forty guests, one exhausted magician, and a present table the size of a small car.” Bill leaned forward, peering through the windshield as if the house might suddenly explode into confetti. The plastic crown he insisted on wearing kept sliding over one eye. Bill wiped sweat from his forehead. “Traktor? They named him after a farm vehicle?” “It’s a very competitive suburb,” Molly said. “Now listen. We walk in as the Party Patrol, a completely fake courtesy service that absolutely does not exist.” Bill blinked. “So we’re lying to children.” “We’re enhancing their wonder,” Molly corrected. “You’ll wear the dinosaur onesie. I’ll carry the clipboard. Nobody questions a woman with a clipboard, especially if she looks busy and angry.” Bill held up the onesie. It was lime green, with a tail that dragged on the floor. “This onesie has a tail. I can’t sneak in a tail. I’ll knock over a punch bowl and start a sticky domino effect.” The tail gave a sad flop, as if it too understood the risks of suburban burglary. Molly refused to acknowledge the flop, because acknowledging the flop would mean acknowledging doubt. “You’re not sneaking,” Molly said. “You’re waddling with purpose. You’re a distraction tank. While every toddler climbs you like a jungle gym, I load the presents into the gift-wrap trolley.” Bill gasped. “The trolley? That’s the sacred vessel. Last month a four-year-old guarded one with a foam sword and pure menace.” “Which is why we replace the loot with oranges,” Molly said. “Nothing says ‘mystery’ like fruit. We roll the trolley to the van, dump the toys, and vanish before the cake is cut.” Outside, toddlers sprinted past, screaming with sugared joy. Bill considered this. “What about the mall cop uncle? The one with the reflective belt?” Molly smiled. “Mall cops must announce themselves before chasing anyone. By the time he yells ‘Stop right there,’ we’ll be three houses away and laughing in the getaway turtle.” Bill strapped on the tail. “Fine. But if a kid asks me to roar, I’m roaring. It’s called staying in character.” Molly started the van. The engine hiccupped, then purred like a cat with a grudge. “Showtime, T-Rex. Let’s go commit toy larceny.” Bill gave a small, sad roar. “Aye, Party Patrol.” They stepped into the sunshine, ready to face sugar, chaos, and the greatest present heist in suburban history.
created_at
"2026-09-03T15:29:16.589Z"
9/3/2026, 3:29:16 PM
model
"tencent/hy4-preview-20260827"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1220
moderation_latency
(null)
generation_time
318355
tokens_prompt
75
tokens_completion
19605
native_tokens_prompt
83
native_tokens_completion
23607
native_tokens_completion_images
(null)
native_tokens_reasoning
22875
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
0.059110329
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788449356-dOGKrp1ZQVdx5iU0g4Gl"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788449356-byH27L77NRox6dc5aLeO"
upstream_id
"6bc8d269-544c-4705-a804-7f8128f0f354"
provider_responses
0
endpoint_id
"ad2a4093-3d2a-41d9-adf1-8b499ac8445d"
id
"6bc8d269-544c-4705-a804-7f8128f0f354"
is_byok
false
latency
1220
model_permaslug
"tencent/hy4-preview-20260827"
provider_name
"Tencent"
status
200
total_cost
0.059110329
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Tencent"
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
100.00%
Dialogue to Total Word Ratio
Ratio: 50.00%, Deviation: 0.00%
neededClean
false
wordsTotal
506
wordsDialogue
253
100.0000%