NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPWFRYJHXGPGE5SZW39QXJ
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
23.1s
Cost:
$0.030315
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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 crisps and ambition. Molly had spread a hand-drawn map across the dashboard, weighted down at one corner by a half-eaten sausage roll. "Right," she said. "Seventeen Larkspur Avenue. Birthday girl is Amelia Fitzwilliam-Hargreaves, turning eight." Bill squinted at the map. "Why does the garden have a moat?" "That's the paddling pool." "You've drawn crocodiles in it." "Artistic licence," Molly said. "Focus. There will be approximately forty presents on a trestle table beneath the gazebo. Wrapped. Ribboned. Unguarded, except by grandparents." "How many grandparents?" "Six, we think. Possibly eight." Bill exhaled slowly, the way a man does when contemplating something enormous. "Eight grandparents, Moll. That's a lot of cardigans." "Cardigans can't run." "Cardigans don't have to run. Cardigans just have to *shout*, and then someone younger runs." He tapped the windscreen. "What about the parents?" "Distracted. There's a bouncy castle arriving at two, and a man doing balloon animals at half past. Between those two events, the adults will be entirely occupied with wondering whether they should have hired the balloon man, because apparently he's been cancelled by another mother on the WhatsApp group." "You've been reading their WhatsApp group?" "I've been reading their WhatsApp group for six weeks. Bill, I know what these people think about the school's new phonics policy. I know things I can never unknow." Bill was quiet for a moment. "And the plan is?" "You go in as the balloon man." "Absolutely not." "You've got the moustache." "The moustache is not a costume, it's a *feature*, and I resent—" "You've got the moustache," Molly repeated, "and I've got a caterer's uniform and a folding trolley. You keep them looking at the balloons. I wheel the presents out through the side gate. Simple." "Can you do a balloon animal?" "I don't have to. You do." "I can do a sausage dog," Bill admitted. "It looks like a sausage. And it looks a bit like a dog. It occupies a haunted middle ground." "Eight-year-olds are a forgiving audience." "Eight-year-olds are the least forgiving audience in the known world," Bill said. "Eight-year-olds will look you in the eye and tell you your sausage dog is disappointing, and then their mother will nod." Molly folded the map with the crisp finality of a woman ending a debate. "Two o'clock. In and out in eleven minutes." "And then what? We sell forty presents we haven't opened?" "That's the beauty of it. It's a lottery. Somewhere in that pile is a games console. Statistically." "Statistically," Bill said, "somewhere in that pile is a bath towel with her name on it, and a book about ponies, and one of those horrible plastic castles that come with two hundred pieces and a warranty nobody can honour." Molly looked at him. "Are you having second thoughts?" "I'm having *first* thoughts, which is more than either of us managed last time." He picked up the sausage roll, examined it, and put it down again. "But no. I'm in." "Good man." "On one condition. We leave one present. The little one. Every kid should get something." Molly considered this. "Fine. But not the games console." "Obviously not the games console," said Bill, wounded. "I'm a thief, Molly, not a saint."
created_at
"2026-07-25T13:19:27.012Z"
7/25/2026, 1:19:27 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2253
moderation_latency
(null)
generation_time
22981
tokens_prompt
75
tokens_completion
840
native_tokens_prompt
93
native_tokens_completion
1194
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
0.030315
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqqkBjXGFm3EhGSFAF1"
is_byok
false
latency
2253
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-1784985567-yWXa7H8DfasSjOUMCbrC"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985567-EzE6lX6WXLvCs2bvuasy"
upstream_id
"msg_011CdNqqkBjXGFm3EhGSFAF1"
total_cost
0.030315
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
0.00%
Matches word count
n/a
neededClean
false
words
538
0.00%
Dialogue to Total Word Ratio
Ratio: 80.80%, Deviation: 30.80%
neededClean
false
wordsTotal
547
wordsDialogue
442
0.0000%