NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KZJ8M7PCB7STMVXEVWEBBS
Run Details
Model:
Hy4 Preview (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
3m 44s
Cost:
$0.040458
Provider:
Tencent
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
99.23%
Dialogue to Total Word Ratio
(details)
99.6170%
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
Molly pressed her face to the living-room window, fogging the glass with an eager puff of breath. Inside, a swarm of sugar-crazed five-year-olds rampaged beneath a banner that read Happy Birthday, Prince Binkles the Corgi. Bill crouched beside her, his ski mask already pulled down despite the fact that it was indoors and seventy-two degrees. “Remind me again why we’re stealing from a dog’s birthday party?” Bill whispered. Molly didn’t look away from the mountain of brightly wrapped packages stacked around the cake. “Because that dog has better taste than you, Bill. I saw the registry. There’s a pony-ride coupon and a solid chocolate rocket ship.” Bill adjusted his mask. “Solid chocolate? At a kids’ party? That’s basically currency. We could trade it for a getaway car.” “Exactly. The parents are in the kitchen refilling punch, and the entertainer quit. The only thing between us and those presents is a guy in a knockoff Spider-Man suit stuck in the bounce house.” “Stuck how?” “Inflated pride, mostly. Plus the door zipper broke. He’s been asking the piñata for help.” Bill nodded solemnly. “Tragic. What’s the plan? Front door or window?” “Window, obviously. You’re going to boost me through that bathroom window, I’ll unlock the back door, and then we sweep in dressed as clowns.” Bill blinked. “Clowns?” Molly pulled two red noses from her bag and tossed him one. “It’s a children’s party, Bill. Nobody questions a clown. They’ll assume we’re part of the entertainment.” “They’ll also assume we’re there to steal presents.” “Only if we act suspicious. Do what I do: balloon animals, weird handshakes, and no sudden moves toward the gift table until the piñata breaks.” Bill tugged the nose onto his ski mask. The effect was horrifying. “I feel ridiculous.” “You look like a birthday miracle. Ready? One—” “Wait,” Bill hissed, grabbing her sleeve. “What if the dog barks?” Molly rolled her eyes. “Binkles is wearing a tiara and eating frosting. He’s not going to bark. He’s going to judge us silently, like a tiny monarch.” “Fine. But if I get glitter on my suit, I’m billing you.” “Noted. Three.” “This is our dumbest plan yet,” Bill muttered. Molly smirked. “Not even close. Remember the lobster disaster?” Bill cupped his hands. Molly stepped up, slipped through the narrow window, and landed with a soft thud beside an inflatable rubber duck. She froze, listening to the distant chaos of squealing children and a grown man sobbing inside a bounce house. Seconds later the back door clicked open. Bill slipped inside, his clown nose glowing faintly in the hallway light. Together they crept toward the laughter, the gifts, and the greatest threat of all: a sticky three-year-old with a firm grip on a juice box, no fear of authority, and a suspiciously adult-looking stare that said she had already memorized both their faces for the police sketch artist. The smell of frosting and chaotic joy hung thick in the air, a sugary fog that made Bill’s ski mask itch again.
created_at
"2026-09-03T15:52:55.694Z"
9/3/2026, 3:52:55 PM
model
"tencent/hy4-preview-20260827"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1207
moderation_latency
(null)
generation_time
223490
tokens_prompt
75
tokens_completion
14967
native_tokens_prompt
83
native_tokens_completion
16149
native_tokens_completion_images
(null)
native_tokens_reasoning
15411
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
0.040457871
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788450775-3nU55nny5lHZwklOezdr"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788450775-l3VGjcwWP0L9DS2JerWX"
upstream_id
"c8fad674-70a7-4639-9fac-6a11a7dd82d0"
provider_responses
0
endpoint_id
"ad2a4093-3d2a-41d9-adf1-8b499ac8445d"
id
"c8fad674-70a7-4639-9fac-6a11a7dd82d0"
is_byok
false
latency
1207
model_permaslug
"tencent/hy4-preview-20260827"
provider_name
"Tencent"
status
200
total_cost
0.040457871
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
99.23%
Dialogue to Total Word Ratio
Ratio: 49.61%, Deviation: 0.39%
neededClean
false
wordsTotal
510
wordsDialogue
253
99.6170%