NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-20-01KHXPQVQ0C76TPWMYF211Y4W6
Run Details
Model:
Mistral Small 3.2 24B
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
3.4s
Cost:
$0.000117
Provider:
Mistral
Evaluation summary
Score
Evaluator
Details Link
0.21%
Matches word count
(details)
87.44%
Dialogue to Total Word Ratio
(details)
43.8275%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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:** "Alright, Bill, listen up. We’re hitting the Thompson kid’s birthday party tonight. Big house, fancy neighborhood, and—get this—his parents are throwing a *fortune* in presents." **Bill:** "A children’s party? That’s your big plan? I’ve stolen from casinos, museums, even a zoo once. A *zoo*, Molly. And now you want me to rob a bunch of five-year-olds?" **Molly:** "It’s not the kids we’re robbing, it’s the *parents*. You ever seen what people buy for their little darlings? High-end gadgets, designer clothes, enough LEGO to build a skyscraper. We walk in, grab the loot, and walk out before anyone notices." **Bill:** "And how exactly do we ‘walk in’? Do I dress as a clown? A magician? A *piñata*?" **Molly:** "No, genius. We go as the *entertainment*. You’re the magician, I’m the face-painter. We distract the kids, you palm the presents, and we’re out before the cake’s even cut." **Bill:** "Face-painter? You think I look like someone who can paint a butterfly on a kid’s cheek?" **Molly:** "You look like someone who just got out of prison. Which, by the way, is exactly why we’re doing this. Now, do you want in or not?" **Bill:** "Fine. But if I have to paint a single unicorn, I’m adding a 20% ‘artistic integrity’ fee to my cut." **Molly:** "Deal. Now let’s go steal some toys."
created_at
"2026-02-20T14:19:57.032Z"
2/20/2026, 2:19:57 PM
model
"mistralai/mistral-small-3.2-24b-instruct-2506"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
198
moderation_latency
(null)
generation_time
3123
tokens_prompt
77
tokens_completion
357
native_tokens_prompt
69
native_tokens_completion
368
native_tokens_completion_images
(null)
native_tokens_reasoning
0
native_tokens_cached
64
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
0.0001173
router
(null)
provider_responses
0
endpoint_id
"23fd6b91-74d8-45fc-ac00-cf141ec7f4a5"
id
"38245751f8ec48a89061aa5a8c8868ba"
is_byok
false
latency
198
model_permaslug
"mistralai/mistral-small-3.2-24b-instruct-2506"
provider_name
"Mistral"
status
200
api_type
"completions"
id
"gen-1771597197-pZsRtQL9fiAfWMyilSCh"
upstream_id
"38245751f8ec48a89061aa5a8c8868ba"
total_cost
0.0001173
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Mistral"
Evaluation details
Result
Evaluator
Details
Meta Data
0.21%
Matches word count
n/a
neededClean
false
words
228
87.44%
Dialogue to Total Word Ratio
Ratio: 96.05%, Deviation: 6.05%
neededClean
false
wordsTotal
228
wordsDialogue
219
43.8275%