NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM62K100NNFTT9N89MZNF4W
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 200 words with 10% dialogue
Duration:
14.0s
Cost:
$0.000326
Provider:
DeepInfra
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 10% 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 and Bill crouched behind a shrubbery, peering through the branches at the brightly lit backyard where a children’s party was in full swing. Streamers fluttered in the breeze, and a piñata swung lazily from a tree branch. The scent of cake and excitement filled the air. Molly adjusted her black beanie, her eyes narrowing as she surveyed the scene. “You seeing what I’m seeing?” Bill whispered, his voice muffled by the scarf wrapped around his face. “Yeah,” Molly muttered. “A logistical nightmare.” Bill pointed at a cluster of gift-wrapped boxes piled near the bouncy castle. “There’s the loot. We grab and go.” Molly shook her head. “Too obvious. They’ll catch us before we even get the first one.” Bill scratched his chin, his brow furrowing. “Clown disguise?” “Worse,” Molly said. “You don’t look good in polka dots.” They sat in silence for a moment, watching as a group of children chased bubbles blown by a frazzled parent. Suddenly, Molly’s eyes lit up. “The piñata,” she said. Bill blinked. “You want to steal candy?” “No,” Molly grinned. “We rig it. Lower it down like a basket, load it with presents, and hoist it back up. They’ll think it’s part of the game.” Bill chuckled. “You’re a genius.” “Obviously,” Molly said, already rummaging through her bag for rope. “Now let’s make this party unforgettable.”
created_at
"2026-02-16T21:35:47.507879+00:00"
2/16/2026, 9:35:47 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
263
moderation_latency
(null)
generation_time
13955
tokens_prompt
77
tokens_completion
388
native_tokens_prompt
66
native_tokens_completion
343
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
0.00032639
router
(null)
provider_responses
0
id
"chatcmpl-RxSvqhPb7uO1hLuUF1EabUdK"
status
200
is_byok
false
latency
25
endpoint_id
"5294d55f-9012-496b-8f22-8cc919432dcd"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277732-OFsA3M6iPJV7qt9mGHNk"
upstream_id
"chatcmpl-RxSvqhPb7uO1hLuUF1EabUdK"
total_cost
0.00032639
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
3.6234%
Matches word count
n/a
neededClean
false
words
224
0.0000%
Dialogue to Total Word Ratio
Ratio: 36.16%, Deviation: 26.16%
neededClean
false
wordsTotal
224
wordsDialogue
81
1.8117%