NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCXTGE42V4T82V7ES8JVP45
Run Details
Model:
Thinking Machines Inkling (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
6m 57s
Cost:
$0.016755
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.05%
Dialogue to Total Word Ratio
(details)
0.0274%
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 overhead light in Molly’s apartment flickered with the enthusiasm of a tired moth, throwing long shadows over the kitchen table where Bill had unfurled a crayon map he’d titled “OPERATION CUPCAKE.” Molly leaned in, her elbow knocking over a jar of dill pickles that served as her paperweight and occasional dinner. “We’re hitting a children’s party,” she announced. Bill looked up, cheese dust painting his chin like a beard. “A party? With music and those tiny sandwiches?” “A birthday party. Twenty-five screaming six-year-olds, one inflatable castle, and a present table that looks like Santa’s warehouse had a yard sale.” Bill frowned. “So we’re robbing kids.” “We’re stealing the presents,” Molly said, tapping the map with her ballpoint pen. “The parents bought them. The parents have salaries. The kids have juice boxes and short attention spans. We have ethics, Bill. We have rules.” “What are they?” Molly held up one finger. “No juice boxes. Two, no crying. Three, we don’t take the cake unless it’s chocolate.” Bill scribbled this on his wrist with a permanent marker. “No juice. No tears. Chocolate contingency only. How do we get inside?” Molly pulled a laminated badge from her coat. “We’re entertainers. I’m Princess Glitterbomb, royal ambassador of sparkle. You’re my security detail, Sparkle-Guard Bill.” Bill straightened, nearly tipping his chair. “Sparkle-Guard. That sounds extremely legitimate.” “It sounds unhinged, but the mother booked us through a website that also sells unlicensed fireworks. We arrive at two, perform twenty minutes of royal small talk, then vanish with the goods while they sing a song about aging.” Bill chewed his thumbnail. “Royal small talk. Is that just talking but with more waving?” “It’s waving with purpose. And the gifts?” Molly traced a circle on the map marked “TREASURE MOUND.” “Stacked under the dinosaur piñata. They open them after the cake, but first they smash the dinosaur with a stick. That’s our window. Candy falls like colorful snow, every parent films on their phone, and we bag the loot.” Bill’s eyes widened behind his glasses. “Candy snow sounds hazardous.” “It’s meteorological sugar, Bill. Stay focused. I have a tiara with a hidden lockpick. You have a cape I made from a thrift-store shower curtain and a helmet covered in aluminum foil.” Bill lifted the foil helmet. It crinkled like a chip bag. “It’s aerodynamic. And intimidating.” “It looks like a baked potato went to war. Put it on.” Bill strapped it on, the peak wobbling. “So we enter as fake royalty, survive the children, steal the presents during the piñata chaos, and escape through the side gate?” “Exactly. The father will be arguing about gluten-free frosting. The grandparents will be debating politics in the corner. The mother will be live-streaming. It’s a criminal’s paradise wrapped in streamers.” Bill saluted, clipping the foil peak against his forehead. “Princess Glitterbomb. I am prepared to serve, sparkle, and steal.” “You are prepared to not mention candy snow,” Molly said, sliding the map into her backpack. “We go in quiet, we come out with plastic treasures, and we leave zero wrapping paper behind.” Bill adjusted his shower-curtain cape, which smelled faintly of mildew and ambition. “What if a child asks for a hug?” Molly paused at the door, her hand on the knob. “Elbow bump, quick compliment about their shoes, keep walking. Sentiment is the enemy of profit.” Bill nodded with grave seriousness. “Elbow bumps. No juice. Avoid the sugar weather.” Molly grabbed her wand-lockpick and the pickle jar for the road. “Let’s make this birthday party legendary for all the wrong reasons.” Bill followed, his foil helmet gleaming in the hallway light. “Wrong reasons are still reasons,” he said, adjusting his cape with theatrical flair. “Let’s go rob childhood.” They stepped into the stairwell, two thieves dressed like a fairy tale that had lost its way, armed with nothing but bad disguises, worse ethics, and a very specific plan to steal joy in its original packaging.
created_at
"2026-07-25T15:20:42.188Z"
7/25/2026, 3:20:42 PM
model
"thinkingmachines/inkling-20260715"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
855
moderation_latency
(null)
generation_time
417247
tokens_prompt
75
tokens_completion
4647
native_tokens_prompt
75
native_tokens_completion
4125
native_tokens_completion_images
(null)
native_tokens_reasoning
3553
native_tokens_cached
32
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
0.01675469
router
(null)
provider_responses
0
endpoint_id
"d6b9d68e-115a-49d7-81cd-1ad1c6942b0e"
id
"chatcmpl-RtkMK1InOXMb4HPkZlwuGQmA"
is_byok
false
latency
96
model_permaslug
"thinkingmachines/inkling-20260715"
provider_name
"DeepInfra"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1784992842-ZayzRjKjztL9yGQaX5VC"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784992842-4Ss6TPvzupuv9oG4t0ZN"
upstream_id
"chatcmpl-RtkMK1InOXMb4HPkZlwuGQmA"
total_cost
0.01675469
cache_discount
0.00002656
upstream_inference_cost
0
provider_name
"DeepInfra"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
656
0.05%
Dialogue to Total Word Ratio
Ratio: 62.26%, Deviation: 12.26%
neededClean
false
wordsTotal
665
wordsDialogue
414
0.0274%