NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KZ4JHYRWK48GBBE41RDV4X
Run Details
Model:
Hy4 Preview (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
4m 24s
Cost:
$0.048466
Provider:
Tencent
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
99.93%
Dialogue to Total Word Ratio
(details)
99.9643%
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 van smelled like old fries and failure. Molly pressed a juice box to the window and peered at the brightly lit house across the street. Balloons bobbed on the mailbox like colorful tumors. Somewhere inside, a herd of sugar-fueled toddlers cheered for a magician who had already quit. Bill sat beside her, struggling into a pair of bright yellow overalls three sizes too small. “Remind me why we’re doing this again?” Bill asked, tugging the straps. “Because birthday presents are basically untraceable cash with extra glitter,” Molly replied. “And because the Woodley wedding wrecked our reputation. We need a win.” Bill sniffed. “A win involving a ball pit.” Molly spread a hand-drawn map across the dashboard. The paper was covered in crayon hearts and arrows. She tapped a large circle labeled LOOT ROOM with a marker that smelled like grapes. “Here’s the plan,” she said. “The gifts go on the dining room table after the cake. We slip in through the side gate, avoid the pony, and grab everything before the piñata explodes.” “What pony?” Bill asked. “The one in the garage. Don’t ask,” Molly said. Bill studied the map as if it were a treasure chart drawn by a deranged preschooler. “And if a kid sees us?” “Then you’re Uncle Steve, the world’s least convincing clown. You do the nose honk and they’ll be too traumatized to follow,” Molly said. Bill held up a red foam nose. “I feel ridiculous.” “Good. Ridiculous is memorable. Incompetent is memorable. We want them telling police about a tall man in tiny overalls, not fingerprints,” Molly said. He strapped on a fanny pack labeled EMERGENCY CAKE SUPPLIES. Molly checked her watch, which was actually a plastic toy watch stolen from a happy meal. The plan was simple, which worried Molly more than a safe full of rabid poodles. She had memorized the guest list, the layout, and the exact moment when parents would be too busy scraping frosting off a tablet to notice two grown adults crawling through a hedge. Bill, meanwhile, practiced his clown walk in the confined space of the van and knocked over a container of juice boxes. The sticky puddle spread across the floor mat like a sugary oil spill. Molly pretended not to notice. She checked the duffel bag: rope, gloves, a spare clown nose, and three bags of trail mix for bribing uncooperative children. Bill pointed at the house, where a cardboard castle now blocked the side gate. “Cake?” Molly shook her head and zipped the bag shut with a decisive hiss. Outside, a chorus of tiny voices began demanding cake again, which meant the distraction window was shrinking fast. “Five minutes,” she said. “Remember the signal?” Bill grinned and squeezed a juice box until it crinkled. “When you squeeze, I scream.” “Not scream. Dance. You’re the decoy, not a banshee,” Molly said. “Same thing,” Bill said. “Showtime?” Molly popped a stick of gum into her mouth and blew a bubble that popped too soon. “Showtime.”
created_at
"2026-09-03T15:45:27.111Z"
9/3/2026, 3:45:27 PM
model
"tencent/hy4-preview-20260827"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4261
moderation_latency
(null)
generation_time
263713
tokens_prompt
75
tokens_completion
17435
native_tokens_prompt
83
native_tokens_completion
19351
native_tokens_completion_images
(null)
native_tokens_reasoning
18644
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
0.048466073
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788450327-7Fixff99rCCIOWqwIrxz"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788450327-3m1gwXDdHKcEQgbaAD9p"
upstream_id
"20674f55-cc28-41bf-a692-7cdb56944338"
provider_responses
0
endpoint_id
"ad2a4093-3d2a-41d9-adf1-8b499ac8445d"
id
"20674f55-cc28-41bf-a692-7cdb56944338"
is_byok
false
latency
4260
model_permaslug
"tencent/hy4-preview-20260827"
provider_name
"Tencent"
status
200
total_cost
0.048466073
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
99.93%
Dialogue to Total Word Ratio
Ratio: 29.88%, Deviation: 0.12%
neededClean
false
wordsTotal
502
wordsDialogue
150
99.9643%