NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM631Q9S2M9EZY83TKSRD94
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 200 words with 50% dialogue
Duration:
13.0s
Cost:
$0.000318
Provider:
DeepInfra
Model input
User/Human
You are a author of comedic heist books. Please write a 200 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 dim glow of Molly’s flashlight flickered across the crumpled blueprint of the neighborhood park. She tapped the paper impatiently with her red-polished nail. "Bill, listen. The piñata’s here, the cake’s here, and the presents—" she jabbed a spot near the picnic table, "—are here. All we need to do is distract the kids, grab the loot, and vanish like we were never there. Easy." Bill scratched his unshaven chin, squinting at the plan. "Distract the kids? With what? A magic show? I can’t even pull a quarter out of someone’s ear without dropping it." "You’re overthinking it," Molly scoffed, rolling her eyes. "We bring a big, shiny balloon animal kit. Kids love that stuff. They’ll swarm you like ants on a lollipop." "And what if they don’t like balloon animals?" Bill countered, nervously twirling the end of his mustache. "What if they’ve got allergies? Or, I don’t know, they hate clowns?" "Fine," Molly snapped, slamming her palm on the table. "You bring candy. Every kid loves candy. Chuck it in the opposite direction, and while they’re scrambling, I’ll grab the presents. Deal?" Bill sighed, slumping in defeat. "Deal. But if one of those little monsters bites me, I’m throwing you under the bus." "Fair enough," Molly grinned, folding the blueprint. "Now let’s go make some kids cry."
created_at
"2026-02-16T21:36:01.508014+00:00"
2/16/2026, 9:36:01 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
291
moderation_latency
(null)
generation_time
12914
tokens_prompt
77
tokens_completion
349
native_tokens_prompt
66
native_tokens_completion
334
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
0.00031838
router
(null)
provider_responses
0
id
"chatcmpl-RSKUs8KBUBCGgGlVV5Tb75N6"
status
200
is_byok
false
latency
48
endpoint_id
"5294d55f-9012-496b-8f22-8cc919432dcd"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277747-BggrOiggxDz6Z7Ole5Qt"
upstream_id
"chatcmpl-RSKUs8KBUBCGgGlVV5Tb75N6"
total_cost
0.00031838
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
20.1897%
Matches word count
n/a
neededClean
false
words
220
84.8749%
Dialogue to Total Word Ratio
Ratio: 56.36%, Deviation: 6.36%
neededClean
false
wordsTotal
220
wordsDialogue
124
52.5323%