NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHY8AGT7QASYC5H6E9RG276C
Run Details
Model:
Ministral 3 3B
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
2.8s
Cost:
$0.000077
Provider:
Mistral
Evaluation summary
Score
Evaluator
Details Link
14.30%
Matches word count
(details)
11.83%
Dialogue to Total Word Ratio
(details)
13.0632%
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 neon glow of the party hall pulsed like a heartbeat, casting jagged shadows across the walls. Molly adjusted her gloves—too tight, but better than nothing—and checked the lock on the back door. Bill leaned against the frame, arms crossed, his breath fogging the air. "You sure this is the right place?" Bill asked, voice low. "I mean, it’s got *kids* in here. What if they’re *too* observant?" Molly rolled her eyes. "They’re kids. They don’t know what ‘observant’ means. Besides, the security system’s got a kid’s security system—like, a *baby monitor* for the front door. Easy pickings." Bill groaned. "You’re right. But what if they *do* notice us? What if they start screaming?" Molly grinned, pulling a small, rusted key from her pocket. "Then we’ll just say we’re *supposed* to be there. ‘Oh, look, it’s the Santa delivery guys!’" Bill groaned again. "You’re insane." "Shut up and help me with this," Molly said, shoving the key into the lock. The door groaned open, revealing a sea of laughter, glitter, and the faint scent of cotton candy. Inside, the room was a riot of color—balloons, streamers, and a giant inflatable unicorn that looked like it had been designed by a very drunk artist. Bill’s eyes widened. "Okay, this is *not* how I pictured a heist." Molly didn’t answer. She was already scanning the room, her fingers twitching toward the nearest gift bag. "We’re going to steal *everything* that’s not nailed down," she declared, her voice echoing off the walls. "And if there’s a single toy that looks like it might be worth more than a candy wrapper, we’re taking it." Bill sighed, following her lead. "Fine. But if we get caught, I’m blaming you." Molly winked. "Deal. Now—who’s got the *real* presents?" Bill groaned. "I *hate* you." The party continued around them, oblivious. Molly and Bill moved like shadows, slipping past parents with too much candy in their mouths and kids who were too busy eating to notice them. The presents were everywhere—under the tree, in the corners, even tucked behind the cake stand. Molly plucked a shiny new bicycle from a pile. "This one’s got *potential*." Bill snatched a stuffed dinosaur from a child’s hands. "And this one’s got *character*." They moved faster now, their hands brushing against gifts, their voices barely audible over the music. The party was a blur of noise and color, but for Molly and Bill, it was just another day at the office. Then, as they reached the back exit, Bill froze. "Wait—what if we get caught?" Molly smirked. "Then we’ll say we were *supposed* to be there." Bill groaned. "You’re *terrible* at this." Molly grabbed his arm. "Shut up and *run*." And with that, they vanished into the night, leaving behind a trail of stolen toys and a party that would never know the truth.
created_at
"2026-02-20T19:27:14.259Z"
2/20/2026, 7:27:14 PM
model
"mistralai/ministral-3b-2512"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
231
moderation_latency
(null)
generation_time
2149
tokens_prompt
77
tokens_completion
755
native_tokens_prompt
69
native_tokens_completion
698
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
0.0000767
router
(null)
provider_responses
0
endpoint_id
"b9f6828f-aef9-498a-91f5-6816fbf72420"
id
"202cde499c7447968229dc36c8bf7b87"
is_byok
false
latency
231
model_permaslug
"mistralai/ministral-3b-2512"
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771615634-MvCFqM1Sb0ZpnwEEJcRu"
upstream_id
"202cde499c7447968229dc36c8bf7b87"
total_cost
0.0000767
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Mistral"
Evaluation details
Result
Evaluator
Details
Meta Data
14.30%
Matches word count
n/a
neededClean
false
words
479
11.83%
Dialogue to Total Word Ratio
Ratio: 36.53%, Deviation: 6.53%
neededClean
false
wordsTotal
479
wordsDialogue
175
13.0632%