NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPKSSCE0T3ESGQ4T2Y9YMC
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
21.2s
Cost:
$0.026165
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
43.38%
Matches word count
(details)
32.98%
Dialogue to Total Word Ratio
(details)
38.1769%
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 van smelled of cold pizza and ambition. Molly had spread a hand-drawn map across the dashboard, weighted at the corners with a stapler, a can of beans, and two of Bill's shoes, which he had removed some forty minutes earlier and never explained. The house across the street was a monument to excess. Balloons in three shades of pink strangled the porch railings. A bouncy castle shaped like a unicorn shuddered in the side yard, sagging slightly at the horn, and a folding table groaned under a mound of gift bags, boxes, and something enormous wrapped in gold foil that Molly had been staring at for eleven minutes. "That one," she said. "That's the one that pays the rent." "Could be a rocking horse." "Could be a pony." Bill leaned forward, squinting through the windshield. His breath fogged a patch of glass, and he wiped it clear with his sleeve, which left it worse. "You've thought about the clown." "There's no clown." "There's a clown, Molly. I saw the van. It said *Chuckles and Company, Weekends and Wakes.*" She folded the map with the precision of a woman who had folded many maps and stolen from all of them. The plan was simple, which was how she liked plans, because complicated plans required trusting Bill with more than one instruction at a time. He would enter through the side gate at 2:15, when the piñata went up. Every adult at a birthday party watched the piñata. It was a biological fact, like blinking, or the way parents drifted toward the wine table in slow, hopeful orbits. Molly would go over the back fence with the duffel bags and work the table from the far end, nearest the hydrangeas, where the sightline from the kitchen window was broken by a wind chime. "Two minutes," she said. "In and out. We take the table, not the tablecloth." "Why would I take the tablecloth?" "Because last time you took the tablecloth, Bill." "It was a nice tablecloth." "There were sausage rolls on it." He had the grace to look out the window. A child in a tiara sprinted past the van, pursued by a smaller child in an identical tiara, both of them shrieking at a frequency that made the beans can vibrate on the dashboard. Somewhere inside the house, a speaker began to play a song about sharing, and Molly felt her jaw tighten in a way she associated with prison food. She checked her watch. Fourteen minutes. She handed Bill a paper hat, which he accepted the way a man accepts a subpoena. "Do I have to?" "You're an uncle. Uncles wear hats." "Whose uncle?" "Nobody's. That's the beauty of it." She pulled the elastic under her own chin and adjusted the brim so it sat at a jaunty, trustworthy angle. "Nobody at a children's party knows who anybody is. Everyone assumes everyone else is related to somebody. It's the most beautiful security failure in the Western world." Bill nodded slowly. Then he tapped the glass. "Molly." "What." "The clown's looking at us." She turned. He was.
created_at
"2026-07-25T13:14:42.613Z"
7/25/2026, 1:14:42 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2318
moderation_latency
(null)
generation_time
20827
tokens_prompt
75
tokens_completion
746
native_tokens_prompt
93
native_tokens_completion
1028
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)
num_fetches
(null)
web_search_engine
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"end_turn"
service_tier
"default"
usage
0.026165
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqUnQqrr14pbMNwdEXb"
is_byok
false
latency
2308
model_permaslug
"anthropic/claude-opus-5-20260723"
provider_name
"Amazon Bedrock"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1784985282-jeF2rQVlaguAPHiCayrK"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985282-7ssZtIK6h8oDUr4uLesg"
upstream_id
"msg_011CdNqUnQqrr14pbMNwdEXb"
total_cost
0.026165
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
43.38%
Matches word count
n/a
neededClean
false
words
517
32.98%
Dialogue to Total Word Ratio
Ratio: 25.29%, Deviation: 4.71%
neededClean
false
wordsTotal
518
wordsDialogue
131
38.1769%