NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHY7Z69V003C9GTMA55DBGQ4
Run Details
Model:
Ministral 3 8B
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
3.3s
Cost:
$0.000102
Provider:
Mistral
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
The neon glow of a flickering *Open* sign cast long shadows across the sidewalk as Molly adjusted the brim of her too-tight beanie, squinting at the three-story mansion ahead. "Alright, Bill," she muttered, "if we’re doing this, we need to move faster than a toddler on a sugar rush. That security guard’s got the reflexes of a goldfish on espresso." Bill, perched on the fire escape like a disgruntled raccoon, adjusted his wire-frame glasses and peered down at her. "Relax, Molly. We’ve got the *perfect* distraction. Remember that guy from the bakery? The one who keeps stealing the last croissant?" "Oh yeah, *Mr. Crumbly*—his name’s actually Gerald, but I refuse to dignify him with his real name." She crouched beside him, voice dropping to a conspiratorial whisper. "We tip him off that the party’s got *extra* goodies—like, the kind of goodies that make you forget your own name. He’ll be too busy licking his fingers to notice us slipping in like ghosts." Bill scratched his chin, leaving a faint trail of lint. "And if he *does* notice? We’ve got a backup plan, right?" Molly grinned, pulling a crumpled flyer from her pocket. "Oh, we’ve got *plans*. Plan A: Gerald gets *very* distracted. Plan B: We use the ‘emergency exit’ sign that’s been stuck on *broken* since 1998. Plan C: We just tell the kids we’re Santa’s elves and they’ll let us in out of sheer terror." Bill groaned. "You’re evil." "Only when it’s necessary." She tapped her temple. "Also, I may or may not have bribed the neighbor’s cat with tuna. If all else fails, *Mr. Whiskers* will lead us straight to the goodies." A distant laugh echoed from inside the mansion, followed by the unmistakable sound of a child screaming—probably from joy, but Molly and Bill both tensed. "Shit," Bill hissed. "That’s the security guard’s laugh. He’s *inside*." Molly’s eyes gleamed. "Then we’re doing this *now*." She grabbed his arm, pulling him toward the side door. "Gerald’s probably already on his way over. We’ve got maybe thirty seconds before he realizes he’s been played." Bill hesitated, then nodded. "Alright, but if we get caught, I’m blaming you for the tuna bribe." Molly winked. "Deal. Now move—before the cat starts judging us." With a final, determined glance at the party lights, they slipped into the shadows, ready to turn a night of presents into their own personal windfall.
created_at
"2026-02-20T19:21:03.044Z"
2/20/2026, 7:21:03 PM
model
"mistralai/ministral-8b-2512"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
224
moderation_latency
(null)
generation_time
2955
tokens_prompt
77
tokens_completion
632
native_tokens_prompt
69
native_tokens_completion
609
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
0.0001017
router
(null)
provider_responses
0
endpoint_id
"c85d26e1-1e0c-449b-9775-2afce7ae510b"
id
"f434ed51ca694303ab5dc4e466e26f35"
is_byok
false
latency
224
model_permaslug
"mistralai/ministral-8b-2512"
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771615263-9lBGtYxqWeoy14GPsgVc"
upstream_id
"f434ed51ca694303ab5dc4e466e26f35"
total_cost
0.0001017
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Mistral"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
404
0.00%
Dialogue to Total Word Ratio
Ratio: 60.15%, Deviation: 30.15%
neededClean
false
wordsTotal
404
wordsDialogue
243
0.0000%