NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5G4X9Q69TSD94GK3M7K88
Run Details
Model:
DeepSeek V3 (2024-12-26)
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
8.6s
Cost:
$0.000411
Provider:
Novita
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
**Molly and Bill’s Great Party Heist** Molly adjusted her oversized clown nose. “You sure this is the right house?” Bill squinted at the invitation. “Yeah, it’s the Andersons’. Kid’s turning six. Cake, balloons, and—most importantly—presents.” “And we’re dressed as clowns because…?” “Because no one questions clowns at a kid’s party. Plus, if we get caught, we can just honk our noses and run.” Molly sighed. “Fine. But if one of those little monsters asks me to make a balloon animal, I’m out.” Bill grinned. “Relax. We grab the gifts, stuff ’em in the giant ‘magic’ sack, and vanish before they sing ‘Happy Birthday.’” “What about the parents?” “Distracted. Always. Half of ’em are on their phones, the other half are pretending to care about toddler gossip.” Molly nodded. “Alright. But if this goes south, I’m blaming you—and your stupid nose.” Bill honked it loudly. “Deal. Now let’s go steal some Legos.” They marched up the driveway, waving at a confused parent. “Party’s this way, right?” Bill called cheerfully. Molly muttered, “We’re definitely going to jail.”
created_at
"2026-02-16T21:25:37.823178+00:00"
2/16/2026, 9:25:37 PM
model
"deepseek/deepseek-chat-v3"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
958
moderation_latency
(null)
generation_time
8411
tokens_prompt
77
tokens_completion
329
native_tokens_prompt
66
native_tokens_completion
296
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
0.0004112
router
(null)
provider_responses
0
id
"0cc124b1bfb1ec8c664f1ba434e679d6"
status
200
is_byok
false
latency
958
endpoint_id
"cefb5a1e-9fea-4496-9a64-4e1b11a7cc8c"
provider_name
"Novita"
model_permaslug
"deepseek/deepseek-chat-v3"
api_type
"completions"
id
"gen-1771277128-RTdlDoHyMbwLKaLtsbEN"
upstream_id
"0cc124b1bfb1ec8c664f1ba434e679d6"
total_cost
0.0004112
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Novita"
Evaluation details
Result
Evaluator
Details
Meta Data
9.6081%
Matches word count
n/a
neededClean
false
words
178
3.2827%
Dialogue to Total Word Ratio
Ratio: 76.40%, Deviation: 13.60%
neededClean
false
wordsTotal
178
wordsDialogue
136
6.4454%