NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHXPBCF2ATNEBXT8X1H1GK4N
Run Details
Model:
Mistral Small 3.2 24B
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
5.9s
Cost:
$0.000174
Provider:
Mistral
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
4.75%
Dialogue to Total Word Ratio
(details)
2.3744%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 adjusted her oversized sunglasses and leaned against the fence of the brightly colored party tent. "You ever think about how weird it is to rob a kids' party?" Bill, balancing a stack of stolen party hats in his arms, shrugged. "Kid parties are the best. No security, no alarms, just a bunch of sugar-high toddlers who won’t remember us. Plus, look at this loot." He gestured to the presents piled under the table. "These parents went all out. I see a drone, a tablet, and—oh, is that a *real* Lego castle?" Molly squinted at the presents. "Yeah, but stealing from kids feels… wrong. Even for us." "Wrong? Molly, we once pickpocketed a nun. A *nun*." "She had a really nice purse." "Exactly! And this is just a bunch of overpriced plastic. Kids don’t even appreciate half this stuff. They’ll play with the box for five minutes and then lose interest." Molly sighed. "Fine. But we’re not taking the stuffed animals. That’s just cruel." Bill rolled his eyes. "Deal. But only because I don’t want to carry them. So, what’s the plan? We just waltz in, grab the presents, and run?" Molly shook her head. "Too obvious. We need a distraction. Something that’ll keep the parents busy while we load up." Bill snapped his fingers. "I got it. You remember that time we crashed that wedding and pretended to be the DJ?" "Yeah, and then we stole the entire buffet." "Exactly! We do the same thing here. You distract the kids, I’ll ‘accidentally’ knock over the cake, and in the chaos, we grab the presents." Molly crossed her arms. "And how exactly do I distract a bunch of five-year-olds?" Bill grinned. "Easy. You tell them you’re a clown." Molly groaned. "I am *not* putting on a red nose." "Come on, it’ll be fun! You can juggle or something." "I can’t juggle." "Then make balloon animals. Kids love that." Molly hesitated. "Fine. But if I have to make a balloon animal, you’re carrying the heavy stuff." Bill clapped his hands. "Deal. Now, let’s go steal some presents before these kids start opening them." Molly adjusted her sunglasses one last time. "You’re a terrible person, Bill." "And you love me for it." He winked and pushed open the tent flap. "Showtime."
created_at
"2026-02-20T14:13:08.395Z"
2/20/2026, 2:13:08 PM
model
"mistralai/mistral-small-3.2-24b-instruct-2506"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
236
moderation_latency
(null)
generation_time
5302
tokens_prompt
77
tokens_completion
592
native_tokens_prompt
69
native_tokens_completion
557
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
0.000174
router
(null)
provider_responses
0
endpoint_id
"23fd6b91-74d8-45fc-ac00-cf141ec7f4a5"
id
"8f35eda42fd24bf2895ab4bcc97d828f"
is_byok
false
latency
236
model_permaslug
"mistralai/mistral-small-3.2-24b-instruct-2506"
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771596788-TWBQSEQhQLwI3JMOTio8"
upstream_id
"8f35eda42fd24bf2895ab4bcc97d828f"
total_cost
0.000174
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
383
4.75%
Dialogue to Total Word Ratio
Ratio: 77.81%, Deviation: 7.81%
neededClean
false
wordsTotal
383
wordsDialogue
298
2.3744%