NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KAYH25CV5C4B4DDNNDB4SH
Run Details
Model:
DeepSeek V4 Flash 0731 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
36.2s
Cost:
$0.000265
Provider:
Wafer
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 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 van idled at the curb, engine a low grumble under the chatter of a dozen sugared-up six-year-olds spilling out of the community center. Bill adjusted his binoculars, focusing on the sagging banner: “HAPPY 7TH BIRTHDAY, JASMINE!” “Okay, Molly,” he said, lowering the lenses. “I see the present table. Fortress of wrapping paper. Two sentries: a mom with a juice box and a dad who looks like he’s already nap-broken.” Molly popped a sour candy, crunching loudly. “Sentinels neutralized. The mom checks her phone every ninety seconds—she’s on a group text about potty training. The dad’s zone of control ends at the coolers. We hit at 2:40, during the hula-hoop contest.” “Hula-hoop contest? That’s not on the intel.” “Checked the party circuit on Instagram. It’s the finale. All kids are in the back yard, and the parents gather to film. The table will be untended for ninety seconds. We’ve got a cover—I’m bringing a gift.” Bill gaped. His mustache twitched. “We’re stealing all the presents. We don’t bring a present.” “It’s a Trojan horse, Billy. A gift. A giant, beautiful, wrapped box. But it’s empty. I set it on the table with the others. When we ‘mistakenly’ grab our box after the party—wait, oh, this is the wrong gift, ha-ha—we take the entire pile in my shopping tote. See? Goodwill gold.” “But they’ll know it’s us.” “No. The box has a big perfect bow. The parent shakes it. It rattles. They set it aside as ‘the one that’s a toy, no, it’s supposed to be surprising.’ Children’s parties are chaos, Bill. They’ve got four uncles, two aunts, a babysitter, and someone’s grandpa crying. They lose everything. Half those gifts go to swap-moms anyway.” Bill rubbed his jaw, glancing at the table again. “How many gifts we looking at?” Molly pulled out a folded receipt. “Estimated thirty-one. We take the top twenty; the rest gets shoved in the cake. The final swipe, I slip, fall, knock over the piñata stick, the big windscreen tears, but I keep the bag.” “You’re going to fall on purpose?” “Bi-political. A sprinkle of tears, ‘’where’s my other sock? I must have dropped it.’ They’ll be so focused on making the kid not cry that we exit with the haul right past the banner.” Bill rubbed his chin. “Where’s the escape route?” “Garage door behind the caterers. Their van is unlocked. Keys under the windshield wiper, I checked. We drive to the pumpkin patch, transfer to your truck, then hide the swag in the back of my aunt’s parakeet store.” “Sounds too smooth,” he grumbled. “You’re a pessimist.” “I’m a professional. How about the dad wails from the stash, and the mom catches us?” “Then you throw the piñata at the mom, and I take the dad’s ankle with my full weight. They won’t press pleads—it’s a kids’ party. The police won’t come ‘cause they need a warrant.” Bill sighed. His finger slid over the table, point to a tiny, painted gift wrapped in glitter. “What about the one small princess princess? I’m feeling a little guilty—that one’s for a little girl.” Molly’s eyes narrowed. “Bill, we took a job. Don’t get soft. Sniffling hearts cause heisting gaps. You promise me—the presents go. The princess gets Sam’s Club mug.” “Can I at least keep the bow?” She slid him a thoughtful look. “You’ll get your bow, love. After we unload, I’ll take you out to a diner. You can keep the cardboard. Just don’t you dare return it pal.” Bill unbuckled his seatbelt, checked his mask. “Fine. But I’m wearing the crusty clown wig. It’s in the spirit.” “As long it don’t bite the movie camera,” Molly said, and they stepped out, two shadows of tinsel and gloom into the laughing sun.
created_at
"2026-09-03T09:52:37.463Z"
9/3/2026, 9:52:37 AM
model
"deepseek/deepseek-v4-flash-20260731"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2313
moderation_latency
(null)
generation_time
36004
tokens_prompt
75
tokens_completion
1050
native_tokens_prompt
89
native_tokens_completion
1024
native_tokens_completion_images
(null)
native_tokens_reasoning
52
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
0.0002649
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788429157-3GkSoX5T4KQSGmvhcW79"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788429157-h4ELzk2dk26yWKntp2ZA"
upstream_id
"16b8d81accd94fd9b08443501a407f61"
provider_responses
0
endpoint_id
"562d5188-97d1-49f7-8881-d0e417e31ada"
is_byok
false
latency
54
model_permaslug
"deepseek/deepseek-v4-flash-20260731"
provider_name
"Makora"
status
429
1
endpoint_id
"c1980fff-0a72-4bce-8d62-a7d3fc582200"
is_byok
false
latency
95
model_permaslug
"deepseek/deepseek-v4-flash-20260731"
provider_name
"DeepInfra"
status
429
2
endpoint_id
"4d5286fc-8f37-40c2-a518-e1ae100d45fc"
id
"16b8d81accd94fd9b08443501a407f61"
is_byok
false
latency
2280
model_permaslug
"deepseek/deepseek-v4-flash-20260731"
provider_name
"Wafer"
status
200
total_cost
0.0002649
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Wafer"
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
632
0.00%
Dialogue to Total Word Ratio
Ratio: 80.81%, Deviation: 30.81%
neededClean
false
wordsTotal
641
wordsDialogue
518
0.0000%