NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KBHQZSX9YF7AHRTJTZXM0E
Run Details
Model:
DeepSeek V4 Pro 0813 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
1m 33s
Cost:
$0.035904
Provider:
Fireworks
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
99.61%
Dialogue to Total Word Ratio
(details)
99.8054%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 30% 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 sat in a stolen ice cream truck parked across the street from a suburban backyard. The truck's jingle still played whenever Bill leaned on the steering wheel, which was often because he had terrible posture and a love of chaos. They watched a children's party unfold through the service window. A bouncy castle shaped like a dinosaur wobbled dangerously every time a child jumped. A face painter applied glitter to a screaming toddler while the mother filmed the tantrum for social media. Presents covered a folding table near the cake, stacked in bright precarious towers, their bows gleaming like bank vault doors. They had been casing the party since the bounce house arrived at dawn. Bill had a notebook filled with crayon diagrams of the present table. Molly had a stopwatch and a growing distrust of Bill's contacts. Molly licked a strawberry cone she had found in the freezer. She pointed at the table. "We go in as entertainers. I'll be Giggles. You'll be Sparkles." Bill adjusted his rainbow wig in the rearview mirror. The wig shed a little rainbow hair onto the dashboard. He checked his reflection and practiced a magical flourish with a wilted carrot. "Sparkles doesn't do manual labor. Sparkles distracts with magic." "That's not a real rule." "It is in the birthday industry." Molly finished her cone and tossed the napkin out the window. It landed in a neighbor's birdbath, startling a plastic flamingo. She didn't care. She was already calculating angles, exit routes, and how many presents she could carry in a canvas sack. "We need a diversion. Something that makes parents look away for ninety seconds." "I hired a pony named Kevin. He has a gambling problem." Molly stared at him. She waited for the punchline. None came. The truck's jingle started again, a tinny rendition of 'Pop Goes the Weasel.' "You can't keep saying that like it's normal." "It explains the raffle ticket incident." "Why is the pony named Kevin?" "He looks like a Kevin. Also he only answers to Kevin." Molly rubbed her temples. "Fine. What about the birthday child? Kids guard gifts like tiny dragons." Bill held up a bubble gun from a duffel bag full of wigs, lanyards, and expired candy. "Birthday Child Distractor 9000. Six hundred bubbles per minute. Tested it on my cat." "And if it fails?" "We dress as party inspectors and confiscate the gifts for safety violations." Molly shook her head. "That's the dumbest plan I've ever heard." "Gluten-free cupcakes." She grabbed the bubble gun. "I'm not wearing face paint. Glitter gives me hives." Bill tossed her a lanyard. "Too late. You're Giggles." Molly looked at the lanyard. It read "Certified Fun Specialist." She hated him. She also cocked the bubble gun and watched a pony named Kevin trot into the yard wearing a tiny party hat. The parents turned, phones raised. The presents sat unattended. Kevin lowered his head and began eating the raffle tickets. The first present slid into Molly's canvas sack.
created_at
"2026-09-03T10:03:07.146Z"
9/3/2026, 10:03:07 AM
model
"deepseek/deepseek-v4-pro-20260813"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
479
moderation_latency
(null)
generation_time
93365
tokens_prompt
75
tokens_completion
8707
native_tokens_prompt
146
native_tokens_completion
9018
native_tokens_completion_images
(null)
native_tokens_reasoning
8349
native_tokens_cached
0
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
0.035904
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788429787-QXtiI2Sz5fecOnVHHYNl"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788429787-jZevLE8rlGVBqHwYfSt7"
upstream_id
"chatcmpl-38b909ac9bc44de095df8308a2c45399"
provider_responses
0
endpoint_id
"ef03a21f-1e70-4f34-9fd9-c6e823f2fcfd"
id
"chatcmpl-38b909ac9bc44de095df8308a2c45399"
is_byok
false
latency
479
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"Fireworks"
status
200
total_cost
0.035904
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Fireworks"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
501
99.61%
Dialogue to Total Word Ratio
Ratio: 30.28%, Deviation: 0.28%
neededClean
false
wordsTotal
502
wordsDialogue
152
99.8054%