NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5CQY8TG9CPF0W6JZCSTJ6
Run Details
Model:
DeepSeek V3 (2025-03-24)
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
16.9s
Cost:
$0.000235
Provider:
DeepInfra
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
**Scene:** Molly adjusted her fake mustache. "Bill, this is the dumbest plan you’ve ever had." Bill grinned. "No, the dumbest was stealing the zoo penguin. This? Genius. Kids’ parties are *lousy* with loot." "You realize we’re robbing a six-year-old’s birthday?" "Exactly! No security, just cake and chaos. We grab the presents, blend in with the clowns—" "You’re not wearing a clown costume." "I’m *incognito*." He tugged his baseball cap lower. Molly sighed. "Fine. But if we get caught, I’m telling the cops you cried during *Frozen*." "Deal. Now, distraction plan: you trip the magician." "Why me?" "Because last time I tripped someone, I *was* the trip." Molly rolled her eyes. "Fine. But I call dibs on the good presents." "Deal. Unless it’s Legos. I love Legos." "You’re a grown man." "And you’re robbing a piñata. Priorities." A beat. Then Molly shrugged. "Fair point. Let’s go steal some toys." Bill pumped his fist. "Operation: Greedy Elf is a go!"
created_at
"2026-02-16T21:23:54.236841+00:00"
2/16/2026, 9:23:54 PM
model
"deepseek/deepseek-chat-v3-0324"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
899
moderation_latency
(null)
generation_time
16846
tokens_prompt
77
tokens_completion
286
native_tokens_prompt
66
native_tokens_completion
291
native_tokens_completion_images
(null)
native_tokens_reasoning
0
native_tokens_cached
32
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
0.00023519
router
(null)
provider_responses
0
id
"chatcmpl-RYqKmU795xiANnLDAzr7UvJ7"
status
200
is_byok
false
latency
56
endpoint_id
"820376cb-f110-4d56-ab52-5bd6ca269420"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3-0324"
api_type
"completions"
id
"gen-1771277017-WwWi8RYj0BhJ5gNEumhD"
upstream_id
"chatcmpl-RYqKmU795xiANnLDAzr7UvJ7"
total_cost
0.00023519
cache_discount
0.00000208
upstream_inference_cost
0
provider_name
"DeepInfra"
Evaluation details
Result
Evaluator
Details
Meta Data
0.0000%
Matches word count
n/a
neededClean
false
words
160
64.6743%
Dialogue to Total Word Ratio
Ratio: 81.88%, Deviation: 8.13%
neededClean
false
wordsTotal
160
wordsDialogue
131
32.3372%