NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHXQC7PW8CGJWJEJ0AS9Q3RW
Run Details
Model:
Mistral Medium 3.1
Test:
Dialogue tags
Scenario:
Write 200 words with 10% dialogue
Duration:
6.4s
Cost:
$0.000660
Provider:
Mistral
Evaluation summary
Score
Evaluator
Details Link
35.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
17.5011%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 10% 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 rainbow-balloon archway sagged under the weight of its own optimism as Molly adjusted her oversized clown bowtie in the reflection of a minivan window. Beside her, Bill fidgeted with a party hat that kept slipping over his eyes, the elastic digging into his stubble like a tiny, judgmental noose. “You realize,” Molly muttered, “we’re about to commit felony grand theft from a room full of sugar-crazed six-year-olds.” Bill adjusted the hat again. “Technically, it’s petty theft. And they’re *distracted* six-year-olds. That’s our window.” A shriek of laughter erupted from the bounce house as a child launched themselves into the air like a human confetti cannon. Molly watched the trajectory with professional detachment. “Distracted is one word for it. ‘Legally unaccountable’ is another.” Bill pulled a walkie-talkie from his fanny pack—stolen from a gym in ’09, still smelling faintly of regret—and cracked his knuckles. “Phase one: blend in. You’re the face painter. I’m the guy who keeps ‘accidentally’ eating the cake samples.” Molly eyed the glitter glue in her pocket. “If I have to turn one more kid into a sparkly raccoon to sell this, I’m taking a cut of the Lego haul.” Bill grinned. “Deal. But if we get caught, you’re the one who ‘misunderstood the dress code.’”
created_at
"2026-02-20T14:31:04.678Z"
2/20/2026, 2:31:04 PM
model
"mistralai/mistral-medium-3.1"
3/1/2001, 12:00:00 AM
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
262
moderation_latency
(null)
generation_time
6350
tokens_prompt
77
tokens_completion
355
native_tokens_prompt
69
native_tokens_completion
316
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
0.0006596
router
(null)
provider_responses
0
endpoint_id
"13c5002b-8c5b-490c-8a5f-52ec0242804f"
id
"5645804498a748ed87d324b0ffd53f0d"
is_byok
false
latency
262
model_permaslug
"mistralai/mistral-medium-3.1"
3/1/2001, 12:00:00 AM
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771597864-ZMdV1gVj0W1NbjX40AQY"
upstream_id
"5645804498a748ed87d324b0ffd53f0d"
total_cost
0.0006596
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Mistral"
Evaluation details
Result
Evaluator
Details
Meta Data
35.00%
Matches word count
n/a
neededClean
false
words
218
0.00%
Dialogue to Total Word Ratio
Ratio: 44.50%, Deviation: 34.50%
neededClean
false
wordsTotal
218
wordsDialogue
97
17.5011%