NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-10-01M25DD74QK6GKJT4GW8TDT23P
Run Details
Model:
DeepSeek V4.1 Flash (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
55.6s
Cost:
$0.008692
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
98.27%
Dialogue to Total Word Ratio
(details)
99.1357%
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 parked the van behind a hedge shaped like a dinosaur and killed the headlights. Through the windshield, a bouncy castle wobbled under a swarm of screaming children. Bill unfolded a crayon map across the dashboard. "Run it again," Molly said. "And if you say clown, I drive away." Bill tapped the map with a cold chicken nugget. "Party starts at two. Chloe turns seven. Twenty-eight kids, six parents, one magician named Marvin the Mediocre. Presents go on the ping-pong table by the patio. Cake at two-forty. That's our window." The map showed the house in crayon, the escape route in glitter, and the dog in a red circle labeled DO NOT TRUST. Molly stared at the nugget, then at him. "Security?" Bill grinned. "One dad filming vertical video, one mum refilling juice, one grandma asleep in a deck chair. The dog is tiny but hostile. He wears a birthday hat." Outside, a unicorn piƱata swung in the breeze like a warning. Molly cracked her knuckles. "And the magician?" "Rabbit trick at two-fifteen. Everyone screams. We slip through the side gate, grab the gifts, and vanish before the cake." Bill traced the route with his finger, smearing nugget grease over the patio. Molly rubbed her temples and checked the van's cracked mirror. "Gifts are in bags. Too bulky. We need pillowcases." Bill produced two princess pillowcases from his jacket. "Already packed. Sparkly camouflage." The pillowcases had sleeping princesses on them, all dreaming of better men. "Nothing says stealth like unicorns." "Also, party hats." "No." Molly looked at the party hats in his lap. They were pink, cone-shaped, and covered in sequins. "With hats, we look like eccentric uncles. Without them, menacing strangers. It's science." "Fine. But no balloons." "Balloons are emergency exfiltration." "Bill." "They float. They distract. They're festive." A balloon escaped the backyard and drifted past the windshield. Bill pointed at it with meaning. Somewhere inside the party, a child screamed with joy or betrayal. The van ticked as it cooled. "Entry point?" Molly unfolded a legal pad covered in jam stains and escape routes. "Side gate. Unlocked. Rusty. It squeaks like a mouse with a grudge." "Alarm system?" "A parrot named Paprika. Says awk and divorce." "Handle her." "I'll offer part of my sausage." "You brought one sausage." "We're stealing presents from a seven-year-old. We can afford two sausages." "Budget is tight. We returned most of last week's loot." "Because the teddy bear had a nanny cam." Molly pinched the bridge of her nose. The van smelled like sausage and bad decisions. "Focus. We take the pink-wrapped presents. Only those." "We're taking all the presents." "Start with pink. Morale." Bill drew a tiny cake on the map and added three candles. "A slice might fall into my hand. That's physics." "Ninety seconds. In and out." "No clowning." Molly pulled on a party hat and started the engine. Bill did the same, grinning like a man who had already eaten cake. "In and out in ninety seconds." "No promises."
created_at
"2026-09-10T10:21:55.739Z"
9/10/2026, 10:21:55 AM
model
"deepseek/deepseek-v4.1-flash-20260910"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
965
moderation_latency
(null)
generation_time
55532
tokens_prompt
75
tokens_completion
12172
native_tokens_prompt
93
native_tokens_completion
14464
native_tokens_completion_images
(null)
native_tokens_reasoning
13753
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
0.00869235
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1789035715-qE0EvVxY9ROd1qpWts3K"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1789035715-V3HkF7qYUxGo84ihGSR0"
upstream_id
"b92c9b98-57d8-4fdd-a344-c3efaa33cb1d"
provider_responses
0
endpoint_id
"fd104bc8-535a-4d31-a847-a8e402e86a02"
id
"b92c9b98-57d8-4fdd-a344-c3efaa33cb1d"
is_byok
false
latency
358
model_permaslug
"deepseek/deepseek-v4.1-flash-20260910"
provider_name
"DeepSeek"
status
200
total_cost
0.00869235
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
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
98.27%
Dialogue to Total Word Ratio
Ratio: 50.59%, Deviation: 0.59%
neededClean
false
wordsTotal
508
wordsDialogue
257
99.1357%