NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2J450N6VR6GJ3F4MS7MTG5
Run Details
Model:
DeepSeek V4 Flash (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
15.5s
Cost:
$0.000287
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.70%
Dialogue to Total Word Ratio
(details)
0.3476%
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
“I’m just saying, Bill, the llamas are a complication I didn’t account for.” Molly squinted through the binoculars at the inflatable bounce house, where a herd of actual llamas was being led past the cotton candy machine. Bill, wedged behind a parked minivan, adjusted his fake mustache. “They’re llamas, not land mines. We walk in, we walk out. The presents are piled by the piñata.” “You’re ignoring the llamas. They have opinions. I saw one spit on a unicorn float. That’s malicious.” “It’s a children’s party, Mol. The theme is ‘Rainbow Safari.’ The llamas are part of the staff. They’re hired.” Molly lowered the binoculars. “Hired llamas. With, what, union contracts? Do they get breaks?” “I don’t know! I don’t care! Focus.” Bill checked his watch. “We have twelve minutes before the cake-cutting. The parents will be distracted, the birthday kid will be crying—standard time window.” “And the llamas will be unsupervised.” Molly tapped her earpiece. “I’m telling you, we need a llama diversion.” “A diversion from a llama is you throwing a bagel at a llama. It’s not a plan.” “Fine. The plan is we wear these.” She produced two brightly colored fanny packs covered in cartoon toucans. “We look like party planners. I say we grab the gift table, shove everything into garbage bags, and walk out singing ‘Happy Birthday’ backwards.” Bill stared at the fanny packs. “Those are horrific. I have standards.” “Your standards are why we’re still broke and sleeping in your mother’s RV. This is a children’s party, Bill. They have juice boxes and optimism. We exploit that.” “We’re not exploiting juice boxes. We’re stealing a three-hundred-dollar drone, a gemstone kit, and probably an iPad. This is a high-net-worth family. Look at the bounce house—it’s two stories.” “It’s also surrounded by llamas. One of them is wearing a party hat, and I think it’s judging me.” Bill sighed. “Okay. New approach. I go in as a balloon artist. I twist balloons, cause chaos. You sneak behind the snack table and bag the loot. We meet at the exit, by the fire hydrant.” “What if they ask for a balloon poodle? Can you do that?” “I can do a snake. It’s a long balloon. Everyone loves a snake.” “Children don’t love snakes. They scream.” “Exactly. Chaos.” Bill grinned. “Then the llamas get spooked, and we have a stampede. Perfect distraction.” Molly nodded slowly. “I hate that it makes sense. But what if the llama with the hat is the ringleader? I’ve seen police dogs. That llama has a vibe.” “We don’t profile livestock, Molly. It’s unethical.” “Profiling llamas is fine. They spit.” A loudspeaker crackled: “Welcome, everyone, to Leo’s Fifth Birthday! Please gather for the pin-the-tail-on-the-llama contest!” Bill froze. “Did she say pin the tail on the llama?” “That is not a game. That is assault.” “Whatever. Go time. Fanny pack on.” Bill strapped on the toucan atrocity. “We do this clean. No llama heroics.” They crept around the hedge. A child screamed. A ballon popped. A llama grumbled. “This is a terrible plan,” Molly whispered. “It’s a fantastic plan. The worst that happens is we get chased by a llama in a party hat.” “That’s the second worst. The worst is we end up in jail, and my mother finds out I was arrested at a five-year-old’s birthday.” “Your mother thinks you’re an accountant.” “I am an accountant! For stolen goods!” “Shut up and grab the presents.” Bill crept toward a cluster of balloons. “On my signal.” “What’s the signal?” “When I say ‘serpent.’” “That’s not a signal, that’s a noun.” “It’s a noun signal. Go!” They moved. The llamas watched. One of them blinked slowly, and Molly knew—this was going to end with her on the evening news.
created_at
"2026-04-25T14:53:58.17Z"
4/25/2026, 2:53:58 PM
model
"deepseek/deepseek-v4-flash-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
703
moderation_latency
(null)
generation_time
15406
tokens_prompt
77
tokens_completion
1120
native_tokens_prompt
67
native_tokens_completion
992
native_tokens_completion_images
(null)
native_tokens_reasoning
44
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
0.00028714
router
(null)
provider_responses
0
endpoint_id
"722e0746-2034-4d6e-a08f-e6271849ee9e"
id
"cd92fd26-e2c2-4c8a-8dfc-1af20e44809b"
is_byok
false
latency
279
model_permaslug
"deepseek/deepseek-v4-flash-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-1777128838-7CXy7ikNhLsa803oPt6g"
session_id
(null)
api_type
"completions"
id
"gen-1777128838-Ud7NgMjb96AG1EqGV0SV"
upstream_id
"cd92fd26-e2c2-4c8a-8dfc-1af20e44809b"
total_cost
0.00028714
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
627
0.70%
Dialogue to Total Word Ratio
Ratio: 79.97%, Deviation: 9.97%
neededClean
false
wordsTotal
639
wordsDialogue
511
0.3476%