NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5HYB0ZP1D470DHKVNCH5B
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
29.9s
Cost:
$0.000723
Provider:
DeepInfra
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 leaned back in the rickety folding chair, tossing a rubber chicken from hand to hand. “Alright, Bill. Let’s hear it. What’s the plan?” Bill adjusted his sunglasses, even though they were indoors. “Simple. We crash the party, blend in with the kiddos, and walk out with the loot. Easy peasy.” Molly stopped tossing the chicken. “Blend in with the kiddos? Bill, you’re six feet tall and have a mustache that looks like it belongs on a cowboy from a spaghetti western.” Bill stroked his mustache thoughtfully. “Fair point. But that’s why we’ve got disguises.” He reached into a duffel bag and pulled out two oversized clown costumes. Molly stared at the costumes, then at Bill. “You want us to dress as clowns? At a children’s party? Have you met children? They’re terrified of clowns.” Bill shrugged. “Exactly. They’ll be too scared to notice us stealing the presents.” “Or,” Molly said, crossing her arms, “they’ll scream bloody murder, and we’ll have to make a run for it with a piñata under one arm.” Bill sighed. “Fine. No clowns. What’s your brilliant idea?” Molly grinned. “Simple. We go as the entertainment.” Bill raised an eyebrow. “Entertainment?” “Yeah. You’ll be the magician, and I’ll be your assistant. We’ll dazzle them with tricks, and while they’re distracted, we’ll swipe the presents.” Bill nodded slowly. “Not bad. But what kind of tricks can we do?” Molly leaned forward, lowering her voice. “Simple stuff. Pulling rabbits out of hats, card tricks, maybe even sawing someone in half.” Bill frowned. “What’s your plan if they actually volunteer someone to be sawed in half?” Molly hesitated. “Uh… we’ll improvise.” Bill groaned. “This is why I’m the brains of the operation.” Molly smirked. “Brains? You thought dressing as a clown was a good idea.” Bill ignored her. “Alright, fine. We’ll go as magicians. But we need a code word for when things go south.” Molly nodded. “How about ‘banana split’?” Bill frowned. “Why ‘banana split’?” “Because if things go south, we split.” Bill sighed. “Fine. ‘Banana split.’ But let’s hope it doesn’t come to that.” Molly shrugged. “With your plan, it’s a coin flip.” Bill glared at her. “This is *our* plan now.” Molly laughed. “Fair enough. So, we’ve got the disguises, the code word, and the distraction. What about the getaway?” Bill reached into the duffel bag again and pulled out a map of the neighborhood. “We park the van two blocks away, make our escape through the back garden, and boom. Clean break.” Molly studied the map. “What’s the plan if someone sees us?” Bill grinned. “We’ll have balloons. Kids love balloons. If anyone sees us, we’ll just hand them a balloon and keep walking.” Molly raised an eyebrow. “And if they’re not buying it?” Bill held up a water gun. “Then we resort to Plan B.” Molly groaned. “Your Plan B is always a water gun.” Bill shrugged. “It’s worked so far.” Molly shook her head. “Alright, magician. Let’s hope you can pull off this trick.” Bill tipped an imaginary hat. “Ladies and gentlemen, prepare to be amazed.”
created_at
"2026-02-16T21:26:58.066192+00:00"
2/16/2026, 9:26:58 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
207
moderation_latency
(null)
generation_time
29884
tokens_prompt
77
tokens_completion
894
native_tokens_prompt
66
native_tokens_completion
789
native_tokens_completion_images
(null)
native_tokens_reasoning
0
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
0.00072333
router
(null)
provider_responses
0
id
"chatcmpl-Rlt0qi3A5gtVAfIP0Kv4qkWf"
status
200
is_byok
false
latency
41
endpoint_id
"5294d55f-9012-496b-8f22-8cc919432dcd"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277187-bUKVagldzl0XyaoHTtGn"
upstream_id
"chatcmpl-Rlt0qi3A5gtVAfIP0Kv4qkWf"
total_cost
0.00072333
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
75.1556%
Matches word count
n/a
neededClean
false
words
513
95.5408%
Dialogue to Total Word Ratio
Ratio: 70.96%, Deviation: 0.96%
neededClean
false
wordsTotal
513
wordsDialogue
364
85.3482%