NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPN778VT0T15NTXRZ7557A
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
21.4s
Cost:
$0.027815
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
3.62%
Matches word count
(details)
0.01%
Dialogue to Total Word Ratio
(details)
1.8153%
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 crayon-drawn map across the table and smoothed it with her palm. "Where did you get this?" Bill asked. "The birthday boy drew it. I asked him where the presents go and he told me. Kids are terrible at operational security." "You interrogated a seven-year-old." "I bought him a popsicle. He volunteered." She tapped the paper. "Presents go on the folding table by the trampoline. Two o'clock, everything's stacked, nobody's watching because that's when the magician arrives." "There's a magician?" "Gus the Magnificent. He does forty minutes and he opens with the rope trick, which means every adult in that yard is looking at Gus, and every kid is looking at Gus, and the presents are looking at nobody." Bill leaned over the map. "What's this square with the teeth?" "That's the dog." "The dog has teeth?" "It's a dog, Bill. That's sort of the arrangement." She sat back. "It's a Pomeranian. It weighs six pounds and it's afraid of ceiling fans." "How do you know it's afraid of ceiling fans?" "The kid told me that too. Popsicle money well spent." Bill rubbed his face. "I want to say something and I want you to hear it without getting defensive." "That's a bad start." "We're stealing presents from children." "From one child." "That's worse. That's so much worse. That's targeted." Molly reached into the bag at her feet and set a glossy catalog on the table, open to a dog-eared page. Bill looked at it. He looked at it for a while. "Is that—" "A limited-run Cascadia Rail set. Two hundred made. Hand-painted. There's one at the bottom of that present pile because Grandpa Reynolds has more money than sense and no idea what a child wants." "How much?" "Fourteen thousand. To the right buyer." Bill sat down slowly. "Okay." "Okay?" "I'm not saying okay to the whole thing. I'm saying okay in the sense that I now understand the shape of the crime." "That's how you always say yes. It's very tiring." "So we take the train set. Just the train set. We leave the rest." "We take the whole table." "Molly." "If we take one box, they know it was a professional. If we take everything, they think it was teenagers." She started rolling up the map. "We drop the rest at the fire station on Corliss. Anonymously. They redistribute at Christmas. The kid gets his stuff back eventually, some of it, and in the meantime he learns that the world is arbitrary and cruel, which is a gift no grandfather can buy." "You've thought about this too much." "I've thought about it exactly the right amount. You're wearing the balloon costume." Bill stopped moving. "The what." "Gus travels with an assistant. Gus's assistant quit in June. I called Gus. Gus is thrilled." "What's the costume?" "It's not technically a balloon." "Molly." "It's a bumblebee, and it's got a zipper in the back that goes all the way down, which means you can carry out about forty pounds of merchandise inside your own thorax." She stood and put on her jacket. "Be at Gus's van at one-fifteen. Don't eat beforehand."
created_at
"2026-07-25T13:15:28.878Z"
7/25/2026, 1:15:28 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1760
moderation_latency
(null)
generation_time
21330
tokens_prompt
75
tokens_completion
779
native_tokens_prompt
93
native_tokens_completion
1094
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
0.027815
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqYC3EpGiSLtK9Yi54x"
is_byok
false
latency
1760
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-1784985328-NAgtlsBr6byd5WiPnXH4"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985328-dIY33U0SvVpsSHEe6fJb"
upstream_id
"msg_011CdNqYC3EpGiSLtK9Yi54x"
total_cost
0.027815
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
3.62%
Matches word count
n/a
neededClean
false
words
524
0.01%
Dialogue to Total Word Ratio
Ratio: 83.80%, Deviation: 13.80%
neededClean
false
wordsTotal
531
wordsDialogue
445
1.8153%