NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KPZVAD4ZS3N3JSZ7H2R8HBFX
Run Details
Model:
Claude Opus 4.7
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
17.0s
Cost:
$0.028015
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
75.16%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
37.5778%
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 booth at Denny's had seen better decades. Molly spread the blueprints across the sticky laminate, knocking over a bottle of ketchup that Bill caught without looking up from his Grand Slam. "So this is the Henderson residence," Molly said, tapping a pen against the paper. "Party starts at two. Cake at three. Present opening at three-thirty." "And we hit at what, three-fifteen?" "Three-twelve. During pin the tail on the donkey." Bill chewed thoughtfully, a bead of syrup quivering on his chin. "Why then?" "Because every kid will be blindfolded and spinning in circles. It's the only window where thirty small witnesses are legally incapacitated." The blueprints, Molly had to admit, were slightly overkill for a ranch-style home in Encino. She'd drawn them herself from memory after casing the place disguised as a Jehovah's Witness. The Hendersons had not wanted to talk about their Lord and Savior, but they had wanted to slam the door, which gave her three full seconds to assess the foyer. "The loot table is here," she continued, circling a spot near the sliding glass doors. "Dora Henderson's little angel turns seven on Saturday, and based on the invitation RSVPs, we're looking at approximately forty-two gifts, average retail value eighty dollars." "That's thirty-three hundred in wrapped merchandise." "Thirty-three hundred and sixty, Bill. Math." Bill set down his fork. "I'm gonna say something, and I don't want you to get weird about it." "Oh no." "These are presents for a child." Molly stared at him. A waitress refilled their coffee and retreated, sensing tension the way horses sense earthquakes. "Bill. We agreed. No more ethics on job nights." "I know, I know, but—" "We stole from a nunnery last month." "That was different. The nuns had a yacht." "A single nun had a yacht. And she was a con woman." "Still." He poked at his hash browns. "A kid's birthday, Molly. That's, like, the Geneva Convention of heists." Molly pinched the bridge of her nose. This was the problem with Bill. He could pick a Medeco lock with a paperclip and a stern look, but he cried at dog food commercials. "Listen to me," she said, leaning in. "Dora Henderson's husband is a hedge fund manager who short-sold his own mother's pension fund. These presents? Tax write-offs from his colleagues. Half of them are regifted from other kids' parties. We're not stealing from a child. We're stealing from the concept of late-stage capitalism." Bill considered this. "Huh." "Besides, we'll leave one. The cheapest one. Symbolic." "Which one's the cheapest?" "Probably whatever comes from the guy who rode over on a bicycle. Scope for bike racks at twelve-fifty." Bill nodded slowly, his conscience successfully laundered. He picked up a crayon from the kids' menu and began sketching escape routes onto a paper placemat featuring a cartoon walrus. "Okay," he said. "But I'm wearing the clown costume this time." "Absolutely not." "Molly." "We'll flip for it."
created_at
"2026-04-24T13:36:57.004Z"
4/24/2026, 1:36:57 PM
model
"anthropic/claude-4.7-opus-20260416"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
632
moderation_latency
(null)
generation_time
16908
tokens_prompt
77
tokens_completion
793
native_tokens_prompt
98
native_tokens_completion
1101
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"
usage
0.028015
router
(null)
provider_responses
0
endpoint_id
"0fa65117-00d9-4835-808d-9b68041d9ce4"
id
"84dd46dd-a8e9-43a4-a590-b96a8c2090dd"
is_byok
false
latency
632
model_permaslug
"anthropic/claude-4.7-opus-20260416"
provider_name
"Amazon Bedrock"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777037817-ltmCBSTzutgvgqcqlyV5"
session_id
(null)
api_type
"completions"
id
"gen-1777037817-py7cDEfSzVKbdypKVLuK"
upstream_id
"84dd46dd-a8e9-43a4-a590-b96a8c2090dd"
total_cost
0.028015
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
75.16%
Matches word count
n/a
neededClean
false
words
487
0.00%
Dialogue to Total Word Ratio
Ratio: 58.43%, Deviation: 28.43%
neededClean
false
wordsTotal
498
wordsDialogue
291
37.5778%