NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHXQ2ZKBSJGTDKYAFNVA5ND6
Run Details
Model:
Mistral Large 3
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
14.7s
Cost:
$0.001094
Provider:
Mistral
Evaluation summary
Score
Evaluator
Details Link
81.27%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
40.6364%
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
The fluorescent lights of the diner buzzed like a swarm of angry bees, casting a sickly glow over the chipped Formica table where Molly and Bill sat. A half-eaten plate of fries congealed between them, forgotten in favor of the blueprint spread across the table. Molly tapped a manicured nail against the paper, her lips pursed in concentration. "You’re telling me we’re hitting a *kids’ party*?" Bill said, rubbing his temples like he was trying to erase the idea from his brain. "I’ve robbed banks, I’ve lifted jewels, I’ve even stolen a *yacht* once—" "Which you crashed into a dock." "—but this? This is where I draw the line. We’re gonna get arrested by a clown." Molly rolled her eyes. "Oh, please. Clowns are terrifying, but they’re not cops. Besides, this is the perfect gig. Low security, high reward. Those parents are gonna be so distracted by their little angels turning into sugar-fueled demons that they won’t even notice us slipping out with a sack full of loot." Bill groaned, slumping back in the booth. "I don’t know, Mol. This feels wrong. Like, morally wrong. We’re stealing from *children*." "First of all, we’re stealing from *parents*," Molly corrected, stabbing a fry into the ketchup for emphasis. "Second, those kids don’t *need* another action figure or a plastic pony. They’ve got enough crap to open their own toy store. And third—" She leaned in, lowering her voice. "The birthday kid’s dad is some tech bro who just dropped twenty grand on a *custom* Lego Death Star. That thing’s worth more than my car." Bill’s eyes widened. "Twenty grand? On *Legos*?" "Yep. And it’s just sitting there, waiting for us." Molly grinned, folding up the blueprint. "We go in as caterers, swap out the real presents with empty boxes, and walk out like we own the place. Easy." Bill sighed, running a hand through his hair. "And what if someone *does* notice? What if some over-caffeinated mom with a phone full of security cam footage sees us and calls the cops?" Molly waved a dismissive hand. "Please. By the time they figure out what happened, we’ll be long gone. And if anyone asks, we just say we’re with the *magician*." "The magician?" "Yeah. You know, the guy who makes the rabbit disappear? We’re just helping with the *illusion*." She smirked. "Besides, if things go south, we can always blame the clown." Bill groaned again, but this time, there was a hint of a smile tugging at his lips. "You’re insane, you know that?" Molly grinned, popping a fry into her mouth. "And yet, you’re still here." "Yeah, well," Bill muttered, grabbing his own fry. "Someone’s gotta keep you from getting us arrested by a guy in face paint." Molly clinked her soda glass against his. "To the easiest score of our lives." Bill hesitated, then clinked back. "To not getting caught by a five-year-old."
created_at
"2026-02-20T14:26:01.462Z"
2/20/2026, 2:26:01 PM
model
"mistralai/mistral-large-2512"
1/1/2512, 12:00:00 AM
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
395
moderation_latency
(null)
generation_time
14667
tokens_prompt
77
tokens_completion
751
native_tokens_prompt
69
native_tokens_completion
706
native_tokens_completion_images
(null)
native_tokens_reasoning
0
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.0010935
router
(null)
provider_responses
0
endpoint_id
"fa718841-07ea-4b7e-b8a3-6069545c7e6a"
id
"a767e2b48413497c83085d764e26dcbf"
is_byok
false
latency
395
model_permaslug
"mistralai/mistral-large-2512"
1/1/2512, 12:00:00 AM
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771597561-TbyaYrNQ2KkvFO1MaPKz"
upstream_id
"a767e2b48413497c83085d764e26dcbf"
total_cost
0.0010935
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Mistral"
Evaluation details
Result
Evaluator
Details
Meta Data
81.27%
Matches word count
n/a
neededClean
false
words
488
0.00%
Dialogue to Total Word Ratio
Ratio: 62.09%, Deviation: 32.09%
neededClean
false
wordsTotal
488
wordsDialogue
303
40.6364%