NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KBCQ2YPCBARGY38EMTXVR9
Run Details
Model:
DeepSeek V4 Flash 0731 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
41.0s
Cost:
$0.000297
Provider:
Together
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
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 through the binoculars, her nose pressed against the grimy window of their stolen panel van. “Okay, Bill, I’m seeing a bouncy castle, a clown with a unicycle, and a table groaning under the weight of wrapped loot. Target rich environment.” Bill, who was trying to pick a wedgie, didn’t look up. “Loot? It’s a five-year-old’s birthday party, Mol. The ‘loot’ is a pile of plastic dinosaurs and a gift card to Build-A-Bear.” “Gift cards are untraceable, Bill. And plastic dinosaurs are the new black on the black market. I know a guy in Jersey who’ll give us forty cents a pound.” “Forty cents a pound? We’d make more money selling our own organs.” “Don’t be dramatic. We’re not here for the money. We’re here for the thrill. And the cake. I’m definitely taking a slice of that cake.” Bill finally looked up, his face a mask of weary resignation. “We are not stealing a child’s birthday cake. That’s a line, Molly. A hard line.” “Fine. No cake. But we’re still hitting the present table. The parents are loaded. I saw the mom get out of a Tesla. A Tesla, Bill. That means the presents are at least Tesla-adjacent.” “Tesla-adjacent? What does that even mean?” “It means there’s probably a remote-control car in there that costs more than our rent. Now, listen. Here’s the plan. You create a diversion. I swoop in, grab the gift bags, and we’re gone before the little monsters can even say ‘pin the tail on the donkey’.” “What kind of diversion? Am I supposed to set the bouncy castle on fire?” “No, that’s too much. Just… pretend to be a concerned dad. Walk up to the clown and tell him his fly is open. That’ll cause a commotion. Clowns are very self-conscious about their pants.” Bill rubbed his temples. “And what are you going to be doing while I’m humiliating a professional entertainer?” “I’ll be the catering lady. I’ve got a spare hairnet and a tray of lukewarm pigs in a blanket. I’ll walk right past the present table, trip, and ‘accidentally’ fall onto the gifts. When I get up, the presents will be in my apron. It’s foolproof.” “It’s not foolproof. It’s a cartoon. You’re describing a cartoon. What if the kids see you? They’re tiny, but they’re vicious. They have no empathy. They’ll scream.” “Kids are easy to confuse. I’ll just tell them I’m a magician’s assistant and the presents are part of the act. They’ll believe anything. They still think the clown’s unicycle is real.” Bill sighed, a long, defeated sound. “I hate this. I hate that I’m even considering this. But the rent is due, and my last job was stealing hubcaps from a Prius. That’s not a career, Molly. That’s a cry for help.” “See? This is our big break. A children’s party. Low security, high sugar content. We’ll be in and out in ten minutes. Now, do you have the hairnet?” Bill reached into his pocket and pulled out a crumpled, slightly sticky hairnet. “I found it in a dumpster behind a diner. It smells like old gravy.” “Perfect. That’s the smell of success. Now, let’s go over the signal. When I give the thumbs-up, you start the clown thing. When I give the thumbs-down, you start screaming ‘fire’ and we abandon ship.” “What if I have to sneeze?” “Then you sneeze, Bill. We’re professionals. We adapt.” Molly slapped her knees and grinned. “Alright, partner. Let’s go steal some happiness.” Bill stared at the hairnet. “I think I’d rather steal the unicycle.”
created_at
"2026-09-03T10:00:22.381Z"
9/3/2026, 10:00:22 AM
model
"deepseek/deepseek-v4-flash-20260731"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
516
moderation_latency
(null)
generation_time
40946
tokens_prompt
75
tokens_completion
1028
native_tokens_prompt
146
native_tokens_completion
987
native_tokens_completion_images
(null)
native_tokens_reasoning
90
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
0.0002968
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788429622-4cXPAeXEFfv3cff61Z2q"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788429622-FwZ7zNi30kLDShERA1qa"
upstream_id
"703f239d-aws_ue2"
provider_responses
0
endpoint_id
"604391ba-7c17-4f8b-8a18-b38c5b45ac52"
id
"703f239d-aws_ue2"
is_byok
false
latency
516
model_permaslug
"deepseek/deepseek-v4-flash-20260731"
provider_name
"Together"
status
200
total_cost
0.0002968
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Together"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
594
0.00%
Dialogue to Total Word Ratio
Ratio: 87.89%, Deviation: 17.89%
neededClean
false
wordsTotal
603
wordsDialogue
530
0.0000%