NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCRBZ89908ZTQB8R7R4HYKC
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
23.1s
Cost:
$0.030415
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
68.10%
Matches word count
(details)
1.51%
Dialogue to Total Word Ratio
(details)
34.8076%
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 spread across the kitchen table were, in fact, a placemat from a pizza restaurant, but Bill had drawn on it with a red marker and that gave the whole operation a certain gravity. He had labeled the bouncy castle "OBSTACLE," the punch bowl "DISTRACTION," and the gift table — circled three times, with arrows — "THE VAULT." Molly examined it while eating cereal directly from the box. "You've drawn the piñata bigger than the house." "The piñata is critical. When that thing bursts, forty children go feral. Total sensory chaos. We walk out through the confusion carrying everything." "And nobody notices two adults in balaclavas hauling a wagon of gifts." "Not balaclavas." Bill produced a paper bag and, with the reverence of a man unveiling a Fabergé egg, withdrew two foam clown noses and a pair of enormous novelty glasses. "Party clowns. Invited guests. Hidden in plain sight." Molly stared at the noses for a long moment, then set down the cereal box, which was the closest she came to expressing respect. The target was the seventh birthday of one Tobias Pemberton-Hyde, whose father owned four car dealerships and whose mother had rented an actual pony. Reconnaissance — conducted by Molly, from a bench, with a newspaper she never turned a page of — indicated a gift table of approximately eleven cubic feet, groaning under boxes wrapped by professionals. There were bows the size of cabbages. There was, allegedly, a remote-control helicopter. "Security?" she asked. "One grandmother, and she's mostly asleep by two. Two caterers, both vaping behind the hedge on rotation. And the pony guy, but the pony guy is not police." "The mother?" "Filming everything for the internet. Which means she's looking at a phone, not at us." Bill tapped the placemat. "We go in at three-fifteen. Cake at three. Sugar peaks at three-twelve. By three-fifteen the birthday boy has bitten someone and everyone's dealing with that." Molly frowned. "You've timed a child's blood sugar." "I've timed *four* children's blood sugar. I sat in a hedge outside a soft-play centre for a week to build that model." "And you told me that was a holiday." "It was restful." The plan required a wagon, a tarpaulin, and a small amount of moral flexibility, of which they had a warehouse. Molly's contribution was the exit: the Pemberton-Hyde garden backed onto a public footpath, and the gate's lock was a padlock so decorative it might as well have been a suggestion. Bill's contribution was the balloon arch, which he intended to collapse at the critical moment because, he said, people always look at falling things. "One rule," Molly said, pulling on a clown nose and immediately regretting it. "We don't take the pony." "I wasn't going to take the pony." "You had a rope in the car." "That rope is multipurpose." "Say it." Bill sighed, the sigh of an artist constrained by philistines. "I will not take the pony." They shook on it. Outside, faintly, a bouncy castle began to inflate, groaning upward like something waking from a long sleep, and Bill checked his watch and smiled.
created_at
"2026-07-25T13:45:22.961Z"
7/25/2026, 1:45:22 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3464
moderation_latency
(null)
generation_time
22925
tokens_prompt
75
tokens_completion
841
native_tokens_prompt
93
native_tokens_completion
1198
native_tokens_completion_images
(null)
native_tokens_reasoning
51
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
0.030415
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNspSJAJQ1geRvemKrUk"
is_byok
false
latency
1260
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-1784987122-Rc40092I9e43q3jOWj4r"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784987122-AFHMZr3tOE20nDOBtZnw"
upstream_id
"msg_011CdNspSJAJQ1geRvemKrUk"
total_cost
0.030415
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
68.10%
Matches word count
n/a
neededClean
false
words
514
1.51%
Dialogue to Total Word Ratio
Ratio: 39.16%, Deviation: 9.16%
neededClean
false
wordsTotal
521
wordsDialogue
204
34.8076%