NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCQTWQ22GJVCDFG9CBXT233
Run Details
Model:
Claude Opus 5 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
22.7s
Cost:
$0.029540
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
62.99%
Dialogue to Total Word Ratio
(details)
61.6347%
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 blueprint spread across the folding table was, technically, a placemat from the Happy Clam Diner, but Bill had drawn on it in three colors of crayon, which he felt lent the operation gravity. Molly leaned over it with her chin in her hand, studying the crooked rectangle he had labeled BOUNCE CASTLE (DANGER). "Explain the danger," she said. "Kids." "That's not danger, Bill. That's ambience." Bill tapped the crayon against his teeth, leaving a waxy blue smear. Outside the garage, rain drummed on the corrugated roof, and somewhere down the block a car alarm had been going off for forty minutes with the persistence of a man who believed in something. Molly had chosen this garage for its privacy, its cheap rent, and the fact that no reasonable person would look for master criminals inside a building that still smelled faintly of dead lawnmower. The target was 14 Fennimore Court. The occasion was the seventh birthday of one Tobias Grimsby-Kane, whose mother had posted the guest list, the schedule, and a photograph of the gift table to a public parenting forum, along with the words *no plastic, please, we're a wooden toy family.* Forty-one children invited. Molly had counted the RSVPs twice. Forty-one children meant forty-one presents, and in a neighborhood where the mailboxes had little copper roofs, forty-one presents meant a van full of merchandise and a comfortable spring. "Timing," Molly said. "Cake comes out at three-fifteen. Everyone goes inside. Candles, singing, the whole ceremony. That's our window." "How long's a window?" "Ninety seconds of singing. Another ninety of a woman named Priya photographing her son's face. Call it four minutes." "Four minutes to move forty-one presents." "Which is why you're going to be a clown." Bill straightened up so fast his knee hit the table. The crayons rolled. He looked at her the way a dog looks at a vacuum cleaner, with a full understanding of the situation and no way to voice objection. "I don't do clown," he said. "You did clown in Tulsa." "That was a *mime*, Molly, and it ruined me." She let it hang. That was the trick with Bill; you laid the idea down like a slice of ham on a counter and waited. He circled it, muttered, poured himself a coffee he wouldn't drink, and eventually came back to the table with the resigned posture of a man selecting his own coffin. "Wig or no wig," he said. "Wig. Big red one. Nobody looks at a clown's face, they look at the hair, and the hair isn't yours." "And the presents?" "Go in the trick trunk. You wheel it out past the parents while doing that thing you do with the balloon animals." "The dachshund." "The dachshund is a masterpiece." Bill nodded slowly, warming, because flattery worked on him like a heat lamp. He picked up the green crayon and drew an arrow from the gift table to the driveway, then added a small stick figure with an enormous head, which Molly understood to be a self-portrait, and which she chose, generously, not to comment on.
created_at
"2026-07-25T13:36:03.306Z"
7/25/2026, 1:36:03 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3785
moderation_latency
(null)
generation_time
22580
tokens_prompt
75
tokens_completion
813
native_tokens_prompt
93
native_tokens_completion
1163
native_tokens_completion_images
(null)
native_tokens_reasoning
52
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
0.02954
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNs7BmttYBzeqPTfZ8Ft"
is_byok
false
latency
1632
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-kgJByLS5c2rgFxCIGDO7"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784986563-o5aGcP4PGSL4b2ethtPY"
upstream_id
"msg_011CdNs7BmttYBzeqPTfZ8Ft"
total_cost
0.02954
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
60.28%
Matches word count
n/a
neededClean
false
words
515
62.99%
Dialogue to Total Word Ratio
Ratio: 26.96%, Deviation: 3.04%
neededClean
false
wordsTotal
523
wordsDialogue
141
61.6347%