NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCVFA4TBDAQY4N3X7EQH0F6
Run Details
Model:
Thinking Machines Inkling (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
30.5s
Cost:
$0.010504
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 30% 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 unfurled the stolen blueprint across the battered folding table with the theatrical gravity of a general declaring war on a piñata factory. Bill sat opposite, chewing a pretzel he’d excavated from the glove compartment of their getaway van, a vehicle he insisted on calling The Regret despite the fact that it was technically a 2003 Honda Civic with a suspicious rattle. "We're hitting a birthday party," Molly announced, tapping a crimson circle drawn around a rented bouncy castle. Bill swallowed hard. "A children's party. You want us to rob toddlers." "Not the toddlers," Molly corrected, adjusting her black turtleneck, which was already damp with sweat despite the air conditioning having died in March and never received a proper burial. "The presents. Every single one. They’re stacked in the corner of the community center like a glittering dragon's hoard. I conducted reconnaissance for three days. Twenty-seven guests, two magicians—one of whom I strongly suspect is running a side hustle in counterfeit baseball cards—and a cake sculpted into a dinosaur that I am not emotionally equipped to discuss." Bill leaned back, pretzel dust cascading onto his cargo pants. "Last Tuesday we lifted a vintage jukebox from a speakeasy. This is downward mobility, Molly. We have standards. We have a reputation." "The jukebox was hollow," she reminded him, sliding a laminated badge across the table. "This is liquid gold. These parents purchase gifts like they're competing in the Olympics of guilt. Electronic tablets, remote-control helicopters, a miniature kitchen set that probably costs more than our collective rent." Bill studied the floor plan, his finger tracing the entrance. "How do we breach the perimeter?" "Costumes," Molly said, producing two lanyards with the confidence of a magician pulling not a rabbit but a felony out of a hat. "We are entertainers. You are Bunches the Clown, and I am Sparkles the Face Painter." Bill held up his badge. "Bunches? That sounds like a gastrointestinal emergency." "It sounds innocent. Innocent bypasses security cameras, suspicious mothers balancing cups of lukewarm chardonnay, and that one dad who definitely thinks he's in a true-crime documentary." Bill scratched his scalp. "What about the dog? The invitation mentioned a therapy golden retriever named Biscuit." "Biscuit receives a strategically placed trail of cocktail sausages," Molly said. "Then he becomes our ally. Meanwhile, you execute the piñata diversion—strike it with the provided bat, scream as though you're being disemboweled, draw every eye in the room. I will bag the merchandise into these recyclable sacks printed with 'World's Best Grandma.'" Bill sighed, the pretzel now a defeated crumb. "This is humiliating." "It's elegant," Molly insisted. "Children's parties possess the worst security architecture in the civilized world because every adult assumes the gravest threat is a sugar rush followed by a nap refusal. We enter, we perform with deliberate incompetence, we exit carrying three garbage bags of pure retail therapy." Bill picked up the neon orange wig, which smelled faintly of cedar blocks and poor life choices. "If I wear this atrocity, I want a bonus." "You may claim the remote-control helicopter," Molly said. "I am reserving the dinosaur cake pedestal for my personal collection." "Agreed," Bill said, pulling the wig over his head. "Terms accepted." Molly checked her watch with the precision of a bomb technician. "We deploy in eighteen minutes. Memorize the extraction protocol." Bill nodded, his face already half-buried in synthetic curls. "Shout pineapple, and we abandon the goods to preserve our freedom. Even career criminals require exit strategies." "Exactly," Molly said, shouldering the sacks. "Let's go steal some manufactured joy." Bill stood, adjusting oversized shoes that squeaked with every step. "The emotional damage is complimentary, I assume?" Molly grinned, adjusting her own face-paint kit. "Always. We don't charge for the trauma. That's just customer service."
created_at
"2026-07-25T14:39:38.545Z"
7/25/2026, 2:39:38 PM
model
"thinkingmachines/inkling-20260715"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
120
moderation_latency
(null)
generation_time
29437
tokens_prompt
75
tokens_completion
3171
native_tokens_prompt
75
native_tokens_completion
2575
native_tokens_completion_images
(null)
native_tokens_reasoning
1733
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
0.01050375
router
(null)
provider_responses
0
endpoint_id
"9911deb9-f44e-4e18-87fe-9d7b60acee72"
id
"otYVGXh-6z2FuN-a20bea4a2ff87816"
is_byok
false
latency
115
model_permaslug
"thinkingmachines/inkling-20260715"
provider_name
"Together"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1784990378-SW1QiUrvv6zbbwMenOjh"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784990378-k2fC37FAqSkoY19MjdOH"
upstream_id
"otYVGXh-6z2FuN-a20bea4a2ff87816"
total_cost
0.01050375
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Together"
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
628
0.00%
Dialogue to Total Word Ratio
Ratio: 61.04%, Deviation: 31.04%
neededClean
false
wordsTotal
634
wordsDialogue
387
0.0000%