NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPM2K96JXYH9MSBA480Z61
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
26.1s
Cost:
$0.028190
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
43.38%
Matches word count
(details)
29.39%
Dialogue to Total Word Ratio
(details)
36.3819%
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
The van smelled of stale popcorn and betrayal, mostly because Bill had eaten Molly's popcorn. "Right," Molly said, spreading a napkin across the dashboard. "Here's the layout. This is the bouncy castle." "That's a coffee ring." "It's a bouncy castle." "It's got a handle." "That's the entrance." Molly stabbed the napkin with a pen. "Presents go on the trestle table, here, under the banner. Fourteen gifts, minimum. Possibly a scooter." "A scooter." Bill sat up. "How much is a scooter?" "Sixty quid. Ninety if it's got lights." "It'll have lights. Kids love lights." He squinted through the windscreen at the community hall, where a woman in a cardigan was wrestling a helium canister. "Molly. I want to say something, and I want you to hear it in the spirit it's meant." "Go on." "This is a child's birthday party." "Correct." "We're stealing from a seven-year-old." "We're stealing from *fourteen* seven-year-olds, Bill, which is arguably more ethical because the loss is distributed." Bill chewed the inside of his cheek. "That's not how ethics works." "You did a term and a half of philosophy at a college that's now a Wetherspoons. You don't get to referee." "I know what a *child* is, Molly." He waved a hand at the hall. "I've got a niece. Last Christmas she opened a jigsaw and cried with joy. Actual tears. Over a jigsaw of a badger." "Then she's easily pleased and will barely notice." "That's worse." Molly turned in her seat. "Bill. Do you remember the marina job?" "I'd rather not." "Do you remember why we're in a van that smells like a cinema floor instead of a villa in Portugal?" "Because of the swans." "Because of the swans," Molly agreed. "So today we go small, we go soft, and we go *unguarded*. Nobody insures a pass-the-parcel." Bill stared at the napkin for a long moment. "What's the plan, then?" "You're going in as the entertainer." "I'm sorry?" "Mrs. Pargeter booked a magician called Marvo. Marvo has, tragically, been double-booked at a christening in Rotherham." Molly reached into the back and produced a purple cape, which had clearly once been a curtain. "You are now Marvo." "I don't know any magic." "You know that thing with the coin." "That's not magic, that's me dropping a coin and looking surprised." "Kids love it. Meanwhile I'm the auntie nobody can place. Every party has one. I hover near the buffet, I say 'hasn't he grown,' and I move the presents out the fire door in a wheelie bin." "There's a wheelie bin?" "There's always a wheelie bin, Bill. That's the beauty of Britain." Bill took the cape. He held it up. It said *SALE* on one corner in dry-cleaning ink. "Fine," he said. "But not the badger jigsaw." "There isn't a badger jigsaw." "If there is one, it stays." Molly considered him. "One jigsaw." "One jigsaw." "Put your cape on, Marvo."
created_at
"2026-07-25T13:14:51.375Z"
7/25/2026, 1:14:51 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
6842
moderation_latency
(null)
generation_time
25964
tokens_prompt
75
tokens_completion
729
native_tokens_prompt
93
native_tokens_completion
1109
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
0.02819
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqVTU6LGFctnVnHj8A1"
is_byok
false
latency
6824
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-1784985291-xCaMDcX3qm6g33vFfajk"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985291-eaNEg9J9WHEMAjbm5Mq2"
upstream_id
"msg_011CdNqVTU6LGFctnVnHj8A1"
total_cost
0.02819
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
483
29.39%
Dialogue to Total Word Ratio
Ratio: 74.95%, Deviation: 4.95%
neededClean
false
wordsTotal
491
wordsDialogue
368
36.3819%