NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM6175QNRM99233P7FMPT9F
Run Details
Model:
DeepSeek V3.1
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
8.6s
Cost:
$0.000324
Provider:
AtlasCloud
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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
"Alright, run me through it one last time," Bill said, shifting his weight on the creaky floorboard. "And skip the part where you call me a philistine for not appreciating the artistry." Molly sighed, pointing a laser pointer at a crayon-drawn map. "Simple. The target is a seven-year-old’s birthday party. The parents are distracted, the security is a single, easily-bribed golden retriever named Sparky. Distraction: you trip over the piñata, creating a candy-based riot. While the tiny humans are in a sugar-fueled frenzy, I slip into the 'VIP lounge'—that's the living room—and extract the assets." "The assets being brightly wrapped boxes containing stuffed animals and puzzles?" "Precisely. High-value, low-risk. Parents overspend, the kids are spoiled. It’s a victimless crime." "Victimless? Molly, we’re stealing from children." "They’ll get more! It’s about the principle. We’re not monsters, Bill. We leave the cake. We’re thieves, not animals." "So what’s our exit strategy?" "You grab the big gift bag holding the ‘best grandpa’ mug. I grab the one with the Lego set. We meet at the getaway vehicle—the ice cream truck you ‘borrowed’." Bill stared at the map. "This is a new low, even for us." Molly grinned. "Bill, in this economy, a new low is just a stepping stone."
created_at
"2026-02-16T21:34:57.173739+00:00"
2/16/2026, 9:34:57 PM
model
"deepseek/deepseek-chat-v3.1"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
812
moderation_latency
(null)
generation_time
8533
tokens_prompt
95
tokens_completion
348
native_tokens_prompt
68
native_tokens_completion
320
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
0.0003244
router
(null)
provider_responses
0
id
"bc3853a20ef64a11bdace632aa075dc0"
status
200
is_byok
false
latency
812
endpoint_id
"0bfe3535-ed02-47ea-a98f-60cc8af4fab2"
provider_name
"AtlasCloud"
model_permaslug
"deepseek/deepseek-chat-v3.1"
api_type
"completions"
id
"gen-1771277688-jovyQ6CVhLjYpKZ6481b"
upstream_id
"bc3853a20ef64a11bdace632aa075dc0"
total_cost
0.0003244
cache_discount
(null)
upstream_inference_cost
0
provider_name
"AtlasCloud"
Evaluation details
Result
Evaluator
Details
Meta Data
43.3783%
Matches word count
n/a
neededClean
false
words
217
99.6448%
Dialogue to Total Word Ratio
Ratio: 87.56%, Deviation: 2.44%
neededClean
false
wordsTotal
217
wordsDialogue
190
71.5116%