NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCQSCAY45759HC7S4KADSES
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
22.4s
Cost:
$0.030640
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
35.00%
Matches word count
(details)
10.70%
Dialogue to Total Word Ratio
(details)
22.8502%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 napkin across the table and smoothed it flat with the side of her hand. "That's the bouncy castle," she said, tapping a blot of gravy. "That's the gift table. And that, right there, is our window." "That's a coffee ring." "That's the *window*, Bill." Bill leaned back and folded his arms. "I want it on the record that I have concerns." "Noted." "Seventeen concerns." "Also noted." "Concern the first," he said, holding up a finger. "It's a child's birthday party." "And?" "And I have a soul, Molly. Small, undernourished, but present." "You stole a wheelchair ramp." "That was a *misunderstanding*. I thought it was a skateboard accessory." He rubbed his face. "Whose party is it?" "Tobias Fenwick-Barr. Turning nine." "Never heard of him." "His father owns four car dealerships and a helicopter, and last year the boy received a Rolex." Bill's arms uncrossed slightly. "A Rolex." "A child's Rolex. Solid gold. With a dinosaur on the face." "That's obscene." "It's *ours* is what it is." He picked up the napkin, examined it, and set it back down. "Right. Fine. Say I'm in. Hypothetically. How do we get past thirty parents?" "Forty." "*Forty* parents, all of whom are holding phones and looking for something to be outraged about—" "That's the beauty of it," Molly said. "Nobody at a children's party knows who anyone else is. Everybody assumes you belong to somebody. You just have to look tired and hold a juice box." "That's it? That's the whole disguise?" "You'll also be wearing a lanyard." "A lanyard." "People trust a lanyard, Bill. A lanyard is a badge for cowards." Bill sighed the sigh of a man who had once wanted to be a dentist. "And the presents? There'll be, what, sixty of them? We can't exactly stuff them down our trousers." "Piñata." "Come again?" "Every party has a piñata," Molly said. "Mine is six feet tall, shaped like a unicorn, and hollow as a politician. We wheel it in, we load it up, we wheel it out. Anyone asks, we say the piñata guy needs to inflate it." "You don't inflate a piñata." "Nobody knows that. Nobody has ever known that." "And if some kid takes a swing at it?" Molly slid a small object across the table. Bill picked it up. "Is this a bicycle bell?" "That's a diversion. You ring it, and you shout, 'The ice cream van's here and it's free.' Forty parents will turn into a stampede. Children will trample their own grandmothers." "That's genuinely evil." "That's *marketing*." Bill turned the bell over in his fingers. Rang it once, experimentally. Winced at how much he enjoyed it. "One condition," he said. "Go on." "The boy keeps something. One present. His actual favourite." Molly stared at him. "Why?" "Because I have to sleep, Molly, and lately I've been dreaming about that ramp." "Fine. One present." "Thank you." "He can keep the socks." Bill rang the bell again. "Saturday, then." "Saturday. Wear something with cartoon characters on it. And practise looking exhausted." "Molly," he said, standing, "I've never had to practise that in my life."
created_at
"2026-07-25T13:35:13.765Z"
7/25/2026, 1:35:13 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2853
moderation_latency
(null)
generation_time
22274
tokens_prompt
75
tokens_completion
834
native_tokens_prompt
93
native_tokens_completion
1207
native_tokens_completion_images
(null)
native_tokens_reasoning
39
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
0.03064
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNs3Ybfn2JR7xJe9GWCs"
is_byok
false
latency
1616
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-1784986513-HYmC7Kik513vsFeROPb5"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784986513-ZC6X9AEal3M4D9pmUV8y"
upstream_id
"msg_011CdNs3Ybfn2JR7xJe9GWCs"
total_cost
0.03064
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
35.00%
Matches word count
n/a
neededClean
false
words
518
10.70%
Dialogue to Total Word Ratio
Ratio: 76.69%, Deviation: 6.69%
neededClean
false
wordsTotal
519
wordsDialogue
398
22.8502%