NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2H3QCEKA3YV3GHNTE1E933
Run Details
Model:
DeepSeek V4 Pro (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
29.9s
Cost:
$0.000831
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
99.98%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
49.9920%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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
The sticky heat of the ventilation shaft was making Bill’s knees ache, a sensation not helped by the relentless, tinny echo of a birthday DJ imploring tiny guests to “floss like a diplodocus” two floors below. He shifted, pressing his ear to the vent grate, while Molly, cross-legged beside him, studied a set of blueprints she’d crayoned onto a stolen pizza box lid. “Remind me again why the grab-bags are the primary target,” Bill whispered, dabbing his forehead with a sleeve. “I saw a fondue set down there that screams black-market resale.” Molly didn’t look up. “Because, Bill, the grab-bags contain the personalized thank-you notes. It’s a treasure map to homes where parents are too exhausted to lock their wine cellars. It’s about building a pipeline, not a one-off snatch.” She traced a purple crayon line. “We enter as Flopsy and Dribbles, the less-popular cousins of a famous rabbit duo.” “Flopsy and Dribbles,” Bill muttered. “I still say that sounds like a bladder condition. My left paw is filled with a state-of-the-art glass cutter, and my tail is a modified vacuum seal for loose glitter. I feel absurd.” “You look even more absurd, which is the point. Nobody suspects a six-foot-two rabbit with a palpable air of despair.” Molly tapped the box. “Phase one: the piñata. The birthday boy, Jeremy, has a documented upper-cut like a hummingbird on espresso. He’ll breach the candy shell on the third swing. That’s our smoke and mirrors.” Bill sighed, the sound muffled by the oversized felt rabbit head in his lap. “And yet, the piñata is filled with ethically sourced, sugar-free lentil puffs. What’s the distraction? A mild moral superiority?” “The distraction is the sugar crash that *won’t* happen. The children, unburdened by a glycemic spike, will become eerily focused. They’ll begin organizing the present table by size and structural integrity. That’s when you, Dribbles, roll out the ‘Magical Gift-Shrinking Machine’—our heavily modified wheelie bin with the words ‘Definitely Magic Bin’ stenciled on the side.” “While I’m peddling garbage illusions, what are you doing?” “I’m Flopsy,” she said, as if that explained the cosmos. “I’ll be facilitating a musical statue meltdown, with a playlist engineered to induce a 94% tantrum rate in the under-seven demographic. ‘Baby Shark’ on a recursive loop, key-shifted down a semitone each time. It’s sonic warfare. The aunties will flee, the dads will weep, and I’ll be hoovering up the Lego sets.” A muffled thump echoed from below, followed by a child’s unnervingly calm voice explaining the second law of thermodynamics to a sobbing adult. Bill blinked. “Alright, I’m in. But when the machine ‘shrinks’ the gifts, there’s a failsafe, right? I don’t want to accidentally dematerialize a Nintendo Switch.” “It’s a bin with a false bottom, Bill. The only thing that dematerializes is evidence.” Molly folded the pizza box, a smile creeping across her face. “Now, let’s review your safe-cracking on the juice-box cooler. I want those organic apple-juice pouches gone before the pass-the-parcel trauma sets in.”
created_at
"2026-04-25T14:36:17.484Z"
4/25/2026, 2:36:17 PM
model
"deepseek/deepseek-v4-pro-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
851
moderation_latency
(null)
generation_time
27996
tokens_prompt
77
tokens_completion
1030
native_tokens_prompt
67
native_tokens_completion
922
native_tokens_completion_images
(null)
native_tokens_reasoning
170
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"
usage
0.000831285
router
(null)
provider_responses
0
endpoint_id
"d084170b-dd99-4e53-9ae1-b63881f3cfbf"
id
"7541c115-2d16-436b-a690-1eb5b97bd909"
is_byok
false
latency
506
model_permaslug
"deepseek/deepseek-v4-pro-20260423"
provider_name
"DeepSeek"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777127777-8fvrRbtc1x9SedEPdhOq"
session_id
(null)
api_type
"completions"
id
"gen-1777127777-OBY9pbF38uBAcaMihm22"
upstream_id
"7541c115-2d16-436b-a690-1eb5b97bd909"
total_cost
0.000831285
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
99.98%
Matches word count
n/a
neededClean
false
words
498
0.00%
Dialogue to Total Word Ratio
Ratio: 72.69%, Deviation: 22.69%
neededClean
false
wordsTotal
520
wordsDialogue
378
49.9920%