NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCQTWGJ2E9D2FFR8518T23M
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
21.7s
Cost:
$0.030940
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
0.21%
Matches word count
(details)
86.30%
Dialogue to Total Word Ratio
(details)
43.2586%
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 blueprints were drawn on the back of a pizza box, which Bill considered a step up from last time, when they had been drawn on Bill. Molly smoothed the cardboard flat across the kitchen table and weighted the corners with two mugs, a stapler, and a tin of sardines that had been in the flat longer than either of them. The layout was a masterpiece of criminal cartography: a lopsided rectangle labeled GARDEN, a smaller rectangle labeled BOUNCY CASTLE (DANGER), and a large X where the presents would be stacked, beneath a trestle table draped in Peppa Pig plastic. "Right," she said, tapping the X with a chewed biro. "Ingress at fourteen hundred hours. That's when the magician arrives. Everyone watches the magician. Nobody watches the loot." Bill studied the diagram with the expression of a man reading a will that had gone against him. He was forty-six years old and had once stolen a forklift, and he did not entirely understand how his life had arrived at Peppa Pig. "It's a seven-year-old's birthday, Moll." "It's forty children," Molly said, "at an average of one present each, minimum retail value fifteen quid, which is six hundred pounds of unregistered, untraceable, gift-wrapped merchandise sitting in a suburban garden guarded by nobody but a woman named Deborah." "How do you know she's called Deborah?" "They're always called Deborah." Bill could not argue with this, and so he selected a different objection, one he had been saving. He pointed out that they did not have an invitation. Molly produced two lanyards from her jacket pocket, each bearing a laminated photograph and the words HAPPY TIMES ENTERTAINMENT LTD. in a font that suggested joy at gunpoint. "You're the balloon man," she said. "I can't do balloons." "Nobody can do balloons. You make a long sausage and tell them it's a giraffe with confidence. Confidence is ninety percent of balloons, Bill, and it's a hundred percent of crime." She went on. There would be a cake, and the cake was the crucial thing, because cake meant candles, and candles meant somebody would dim the kitchen lights and sing, and singing meant thirty-one seconds of collective adult distraction. Thirty-one seconds was enough. She had timed it in the bath. Bill asked what he was meant to do during the thirty-one seconds. Molly said he was meant to be reversing a van up the side alley with the engine already running and the back doors already open and his mouth already shut, and that if he sang along she would end him. "And the presents just—what, walk themselves into the van?" "Bin bags," said Molly. "Two bin bags, one sweep, out the gate. Same as Christmas. Better than Christmas, actually, because at Christmas the presents are spread over four rooms and there's a grandmother." There was a long silence, filled by the fridge. "I don't feel good about this," Bill said. "You never do. You'll feel wonderful by Tuesday." Bill looked at the sardines, and at the lanyard, and at the balloon-shaped hole where his dignity had been, and reached for the pizza box to see whether the crusts were still under it.
created_at
"2026-07-25T13:36:03.099Z"
7/25/2026, 1:36:03 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
4476
moderation_latency
(null)
generation_time
21583
tokens_prompt
75
tokens_completion
841
native_tokens_prompt
93
native_tokens_completion
1219
native_tokens_completion_images
(null)
native_tokens_reasoning
56
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
0.03094
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNs7AP4BMbj7Yy5Wf1LZ"
is_byok
false
latency
1434
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-1784986563-jxkrHA7iPKSC6YqlUpfT"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784986563-kquiCCJlXyuNwirPFeyI"
upstream_id
"msg_011CdNs7AP4BMbj7Yy5Wf1LZ"
total_cost
0.03094
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
0.21%
Matches word count
n/a
neededClean
false
words
528
86.30%
Dialogue to Total Word Ratio
Ratio: 31.72%, Deviation: 1.72%
neededClean
false
wordsTotal
536
wordsDialogue
170
43.2586%