NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KB4C18XZGYP9ZFHVRWMJJE
Run Details
Model:
DeepSeek V4 Pro 0813 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
49.4s
Cost:
$0.015666
Provider:
Sail Research
Evaluation summary
Score
Evaluator
Details Link
35.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
17.5013%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 squinted at the crayon-drawn map on the napkin. "Okay, so the bounce house is the primary distraction. You go in as the balloon guy, I'll be the face painter." Bill frowned at his fake mustache. "You said I could be the magician." "Bill, you can't do magic." "I can pull a quarter out of a kid's ear." "That's not magic, that's assault. You almost got arrested at the petting zoo." "That goat had it coming." Bill sighed. "Fine, balloon guy. What's the plan for the presents?" "We wait until the birthday boy blows out the candles. All eyes on the cake. You start twisting balloons into a giraffe, I'll wheel the gift table into the supply closet." "Wheel? It's a folding table, Molly. It doesn't have wheels." "I brought a dolly." "You brought a dolly to a children's party?" "I'm a professional, Bill. It's under my skirt." Bill stared. "How?" "Don't ask questions you don't want answered." Molly tapped the napkin. "Now, the presents are wrapped in shiny paper. We need to move fast, but not so fast that we look like thieves. We're just two exhausted entertainers tidying up." "What if a kid sees us?" "I'll tell them the gifts are going to a secret surprise parade and they'll come back filled with candy." Bill nodded slowly. "That's good. Kids love lies." "They love candy. Lying is just the delivery system." Molly pulled a clown wig from her bag. "You wear this." Bill recoiled. "I'm not wearing a clown wig. Clowns are terrifying." "That's why you wear it. Nobody makes eye contact with a clown. It's the perfect invisibility cloak." "Can't I be a pirate balloon guy?" "At a princess party? Think." Bill grumbled. "Fine. But I'm not painting my face." "You don't have to. Your natural face is suspicious enough." "Thanks." "Welcome. Now, the cake cutting is at two. We arrive at one-forty-five, set up, and smile like we're happy to be there." "What if the parents ask for references?" Molly handed him a business card reading "Twisty Bill's Balloon Wonders." "You have a five-star rating on a website I made last night." Bill squinted. "It says I specialize in balloon animals for funerals." "That's for sympathy balloons. It gives you range." "Who hires a balloon guy for a funeral?" "Grieving people with children, Bill. Think outside the bounce house." Molly checked her watch. "Any more questions or can we go steal some princess castles and robot dinosaurs?" Bill adjusted his mustache. "One last thing. If we get caught, do we run or cry?" "Both. You cry, I run. Tears slow down security." "Why do I have to cry?" "Because you're wearing the clown wig. Nobody chases a crying clown. It's too sad." Bill sighed and started the van. "I hate that you're always right." "Me too, Bill. Me too." Bill put the van in reverse. "Should we bring the spare clown shoes?" Molly shook her head. "No. Clown shoes squeak. We need silence, not comedy." Bill frowned. "What if the kids ask why I'm not funny?" "Tell them you're a mime. Mimes don't talk."
created_at
"2026-09-03T09:55:48.92Z"
9/3/2026, 9:55:48 AM
model
"deepseek/deepseek-v4-pro-20260813"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
566
moderation_latency
(null)
generation_time
49317
tokens_prompt
75
tokens_completion
3574
native_tokens_prompt
146
native_tokens_completion
3954
native_tokens_completion_images
(null)
native_tokens_reasoning
3221
native_tokens_cached
145
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
service_tier
(null)
usage
0.01566554
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788429348-f4fngVLMKMGhNcZb64He"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788429348-dKFZWh2wIGIlszKGj4N0"
upstream_id
"asapreq_2ba62bc54b71784b7aac689c9f67554f"
provider_responses
0
endpoint_id
"6811c06b-8624-4796-ad6e-807e33b97ae6"
id
"asapreq_2ba62bc54b71784b7aac689c9f67554f"
is_byok
false
latency
566
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"Sail Research"
status
200
total_cost
0.01566554
cache_discount
0.00018502
upstream_inference_cost
0
provider_name
"Sail Research"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
35.00%
Matches word count
n/a
neededClean
false
words
518
0.00%
Dialogue to Total Word Ratio
Ratio: 85.82%, Deviation: 15.82%
neededClean
false
wordsTotal
522
wordsDialogue
448
17.5013%