NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPRHQRGMJ1JQPBB2NNGAEW
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
21.9s
Cost:
$0.027990
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
0.03%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0152%
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 van smelled of stale birthday cake, which was Bill's fault, and he knew it. "You ate the reconnaissance cake," Molly said. "I ate half the reconnaissance cake. The other half is evidence." She spread the blueprint across the dashboard, smoothing it with the flat of her hand. It was not, strictly speaking, a blueprint. It was a paper placemat from the Golden Dragon Buffet with a maze on one side, and Bill had drawn the layout of the Hargrove residence in crayon over the top of the maze, so that the front hall now shared space with a cartoon dragon holding a fortune cookie. "Explain this to me one more time," Molly said. "Right. So." Bill tapped the crayon. "Bouncy castle in the back garden, that's the diversion. Twelve children go into a bouncy castle, nobody comes out for forty minutes. It's like a Bermuda Triangle for six-year-olds. Meanwhile the presents are stacked on the trestle table in the conservatory, unguarded, gift-wrapped, portable, and — this is the beautiful part — completely untraceable, because nobody keeps receipts for a toy giraffe." Molly stared at the placemat. She was forty-one years old and had once relieved a Belgian shipping magnate of a Vermeer. "We are stealing from a child," she said. "We are stealing from twelve children, technically, since they brought the presents." "That's worse, Bill. You've made it worse." He held up a finger, which was his gesture for a point he considered devastating. "Tarquin Hargrove is seven years old and his parents hired a magician who was on television. The child has a pony. The child has a pony *and* a trampoline, and last year he got a second pony because he didn't like the face of the first one." "He didn't like its face." "Said it looked judgmental." Bill sniffed. "Frankly, the pony was right." Molly looked out the windscreen at the house, which was the sort of house that had an intercom at the gate and a gravel drive designed for the sound it made. Balloons were tied to the railings in clusters of three, in colors that had been chosen by someone with an interior designer. "Entry," she said. "Side gate, during the piñata." "Timing." "Fourteen minutes, which is how long a piñata lasts when the dad insists on holding the rope." "Extraction." "Bin bags. Two each. We go out through the hedge into Fennel Lane, into the van, gone before the candles." Molly folded the placemat carefully along its original creases and slid it into her jacket. Somewhere beyond the gate, a bouncy castle inflated with the long asthmatic groan of a thing waking up against its will, and a woman's voice shouted something about shoes. "One condition," Molly said. Bill waited. "We leave the birthday boy's present. The one from his gran. There's always one wrapped badly — that's hers." "That's sentimental." "That's professional. Nobody investigates a theft where the gran's gift survives. It reads as a miracle. Miracles don't get fingerprinted." Bill considered this, then nodded slowly, the way men do when they have been outmaneuvered and intend to pretend they agreed all along. "Bin bags," he said, "are in the back, under the cake."
created_at
"2026-07-25T13:17:17.952Z"
7/25/2026, 1:17:17 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1718
moderation_latency
(null)
generation_time
21754
tokens_prompt
75
tokens_completion
803
native_tokens_prompt
93
native_tokens_completion
1101
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
0.02799
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqgDyYTyEPWjJLz8Nc5"
is_byok
false
latency
1705
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-1784985437-J0cyMewY3Iqk75hGDaFf"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985437-Rh6L9XMPqfmUTlDHzl6i"
upstream_id
"msg_011CdNqgDyYTyEPWjJLz8Nc5"
total_cost
0.02799
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
0.03%
Matches word count
n/a
neededClean
false
words
530
0.00%
Dialogue to Total Word Ratio
Ratio: 50.56%, Deviation: 20.56%
neededClean
false
wordsTotal
534
wordsDialogue
270
0.0152%