NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCRJVHN9F976RYNRF8D024E
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
23.2s
Cost:
$0.030890
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
97.61%
Dialogue to Total Word Ratio
(details)
78.9446%
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 blueprint across the hood of the van, smoothing it with the reverence of a surgeon. "That's a place mat," Bill said. "It's a schematic." "It's a place mat from the Sizzler. There's a gravy stain on the west wing." "That's not gravy, that's the bouncy castle." Molly tapped it twice. "Perimeter's soft. Two exits, one gate, and a woman named Deborah who thinks she's running a tight ship." "Deborah being—" "The mother. Forty-one. Runs a book club, hates surprises, allergic to almonds. She'll be manning the cake table from three until three-forty." Bill sucked his teeth. "And the presents?" "Sunroom. Northeast corner. Stacked like the Ark of the Covenant and guarded by absolutely nobody, because Deborah believes in the fundamental honesty of children." "Deborah's never met children." "Deborah's had four." "Then Deborah's a slow learner." Bill peered at the place mat, then at the house across the street, where somebody had tied so many balloons to the mailbox it had begun to lean hopefully skyward. "Talk me through the entry." "You're the entertainment." "I'm sorry?" "You're Pockets the Clown. I booked you at eleven this morning under a fake name and a real deposit. Forty pounds, non-refundable, which means Deborah is now emotionally invested in your success." "Molly." "You do twenty minutes of balloon animals, you work the room, you get the parents laughing—" "Molly." "—and while everyone's watching you drop a plate spinner on your own foot, I go in through the sunroom with three bin bags and a hand truck." Bill straightened up very slowly, the way a man does when he's decided to be dignified about something. "I'm allergic to latex." "You're allergic to effort." "I have a *condition*." "You have a *costume*. It's in the back, next to the hand truck. There's a wig. It's orange." "Is there a nose?" "There's a nose." "Does the nose honk?" "The nose honks, Bill." There was a pause in which a very small argument was won and lost entirely without words. "Fine," Bill said. "But I'm not doing the voice." "Nobody asked for a voice." "Because if I do the voice, that's a character, and if it's a character, that's a witness description. 'Officer, he had a squeaky little voice and he made me a giraffe.' No. I'll be a silent clown. A tragic figure. Very European." "Do whatever gets you through it." Molly folded the place mat and tucked it into her jacket. "Rules. No touching anything with a bow on it under fifty quid—we're not animals. Nothing from the grandparents pile, that's all socks and disappointment. And if a child speaks to you directly—" "I know." "Say it." "I don't engage. I mime sadness and I move away." "Beautiful." Molly checked her watch. "Twelve minutes. Go be beloved." Bill hauled open the van's back door, and stood there staring at a rumpled orange wig with something like grief. "Molly." "What." "If this goes wrong, I want you to remember that I loved you." "If this goes wrong," Molly said, "I'll tell them you came with the balloons."
created_at
"2026-07-25T13:49:08.541Z"
7/25/2026, 1:49:08 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3468
moderation_latency
(null)
generation_time
22762
tokens_prompt
75
tokens_completion
853
native_tokens_prompt
93
native_tokens_completion
1217
native_tokens_completion_images
(null)
native_tokens_reasoning
68
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
0.03089
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNt75xWtJwhJTnnyD8QV"
is_byok
false
latency
1149
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-1784987348-PkqHZzIbT4NIvA8UHdzJ"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784987348-UqcsikBXw05OuxNRUfPH"
upstream_id
"msg_011CdNt75xWtJwhJTnnyD8QV"
total_cost
0.03089
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
97.61%
Dialogue to Total Word Ratio
Ratio: 69.31%, Deviation: 0.69%
neededClean
false
wordsTotal
518
wordsDialogue
359
78.9446%