NC Bench
Overview
About
Tests
Models
Vendors
Compare
Run: 2026-02-24-01KJ8VB3YTK3C4H8XVBF4M19KX
Run Details
Model:
Claude Opus 4.6 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
2m 16s
Cost:
$0.328485
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
9.61%
Matches word count
(details)
16.53%
Dialogue to Total Word Ratio
(details)
13.0690%
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
Molly spread the hand-drawn map across the hood of their dented minivan, weighing down one corner with a stolen juice box and the other with a half-eaten granola bar. The map, rendered in colored pencil on the back of a Denny's placemat, depicted the Hendersons' backyard in extraordinary tactical detail. She had even drawn little stick figures to represent the partygoers, each wearing a cone-shaped hat. "We're looking at fifteen hostiles, ages four through seven," she said, tapping the map with a bitten fingernail. "One of them bites." Bill leaned forward, squinting. He was a large man who looked like he'd been designed for moving refrigerators, not creeping through suburban backyards. His balaclava was already on, rolled up to his forehead like a beanie, which made him look less like a criminal mastermind and more like a confused longshoreman. "Which one bites?" "The birthday boy. Kevin. Got sent home from preschool twice last month for it." Bill nodded slowly, the way he always did when pretending to follow a plan he'd already lost track of. Molly recognized the glassy eyes, the familiar slack in his jaw, and pressed on anyway. The presents, she explained, would be piled on a folding table near the bounce house, roughly twelve feet from the back gate. She had already tested the latch and confirmed it was so flimsy a housecat could pop it open. Their window would come during the cake ceremony, when every child and parent would be huddled around the picnic table watching Kevin try to blow out six candles with the lung capacity of a small accordion. "What if it's ice cream cake?" Bill said. "Those candles go fast. One good breath and they're out." Molly paused. She stared at the map as though an answer might crawl out from between the crayon lines. "We'll have to hope it's buttercream. Buttercream is denser. The candles sink in a little. Buys us time." Bill accepted this with the solemnity of a man receiving a difficult prognosis. He pulled the balaclava down over his face and immediately fogged up his glasses. "I can't see a thing," he said. "Did you cut eyeholes?" "I forgot the scissors. Got distracted by that penguin documentary." Molly closed her eyes and drew a long breath. Down the street she could hear children screaming with joy, the thump of pop music from a portable speaker, and the hiss of a bounce house pump working overtime. The party was underway. She reached into the glove box, retrieved a pair of nail clippers, and tossed them at Bill's chest. "Improvise," she said. "That's what professionals do." "Are we professionals? Last week you reversed into a mailbox fleeing a lemonade stand." "That mailbox was in a blind spot. And we cleared eleven dollars, so I call that a win." Bill began carefully clipping at the knit fabric over his left eye. Molly folded the map, tucked it into her fanny pack, and cracked her knuckles. Somewhere behind that flimsy gate sat a folding table piled with presents wrapped in dinosaur paper and glitter ribbon, and every last one was about to become hers.
created_at
"2026-02-24T22:12:01.121Z"
2/24/2026, 10:12:01 PM
model
"anthropic/claude-4.6-opus-20260205"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1601
moderation_latency
242
generation_time
135745
tokens_prompt
77
tokens_completion
2304
native_tokens_prompt
92
native_tokens_completion
13121
native_tokens_completion_images
(null)
native_tokens_reasoning
1509
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
0.328485
router
(null)
provider_responses
0
endpoint_id
"a8708fb9-08ea-43bb-a3b1-165c87ef09df"
id
"b5658894-f586-44df-a485-da860175501e"
is_byok
false
latency
1579
model_permaslug
"anthropic/claude-4.6-opus-20260205"
provider_name
"Amazon Bedrock"
status
200
api_type
"completions"
id
"gen-1771971121-X02ecT2eXn4L7GH1Kotm"
upstream_id
"b5658894-f586-44df-a485-da860175501e"
total_cost
0.328485
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
Evaluation details
Result
Evaluator
Details
Meta Data
9.61%
Matches word count
n/a
neededClean
false
words
522
16.53%
Dialogue to Total Word Ratio
Ratio: 24.00%, Deviation: 6.00%
neededClean
false
wordsTotal
525
wordsDialogue
126
13.0690%