NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-02-08-01KGYYHG2M5VP3HBVQ5XDEKZWR
Run Details
Model:
Claude Opus 4.5
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
9.2s
Cost:
$0.008675
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
27.17%
Matches word count
(details)
99.72%
Dialogue to Total Word Ratio
(details)
63.4413%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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 crayon-drawn map across the sticky diner table. "So the target is Timmy's seventh birthday party. Security is tight." "How tight are we talking?" Bill leaned forward. "One grandmother at the front door. She's got cataracts but hearing like a bat." "Lovely. What about the presents?" "Stacked in the living room. Behind the bounce house." "There's a bounce house inside?" "These people have money, Bill. That's why we're hitting them." "Fair enough. Entry point?" "Backyard gate. We go in dressed as clowns." "I hate clowns." "Everyone hates clowns. That's why nobody questions them. They just want them gone." "That's actually brilliant." "I know. Now, the tricky part is the dog." "There's a dog?" "Golden retriever. Name's Biscuit." "Aggressive?" "Aggressively friendly. He'll blow our cover with love." "I'll bring treats." "Bring many treats. This dog is a bottomless pit." "What's our exit strategy?" "The magician arrives at three. During his act, everyone watches. We load the van." "We have a van?" "We have my nephew's minivan. It has juice stains and smells like failure." "Perfect cover." "That's what I said."
created_at
"2026-02-08T15:39:59.92476+00:00"
2/8/2026, 3:39:59 PM
model
"anthropic/claude-4.5-opus-20251124"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1822
moderation_latency
202
generation_time
8952
tokens_prompt
61
tokens_completion
274
native_tokens_prompt
75
native_tokens_completion
332
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)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.008675
router
(null)
provider_responses
0
id
"be39ffa2-f572-4f6c-8f7b-1c21a9d7a883"
status
200
is_byok
false
latency
1822
endpoint_id
"de32581c-2e11-4f54-8d29-e8a6e19d3038"
provider_name
"Amazon Bedrock"
model_permaslug
"anthropic/claude-4.5-opus-20251124"
api_type
"completions"
id
"gen-1770565189-cWSqATsVwhvRK4Z0BPsC"
upstream_id
"be39ffa2-f572-4f6c-8f7b-1c21a9d7a883"
total_cost
0.008675
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
Evaluation details
Result
Evaluator
Details
Meta Data
27.17%
Matches word count
n/a
neededClean
false
words
181
99.72%
Dialogue to Total Word Ratio
Ratio: 92.31%, Deviation: 2.31%
neededClean
false
wordsTotal
182
wordsDialogue
168
63.4413%