NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCPP0X302SY71XWX8JVTSTK
Run Details
Model:
Claude Opus 5
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
25.0s
Cost:
$0.029165
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
81.27%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
40.6364%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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 blueprints across the kitchen table, smoothing them with the reverence of a surgeon laying out instruments. Bill leaned over her shoulder, chewing a pen cap. "That's a birthday party," he said. "That's a *fortress*." "It's a bouncy castle and a folding table." Molly tapped the paper. "The bouncy castle is the perimeter. The folding table is the vault. Twelve presents, Bill. Twelve. Unopened, unaccounted for, and guarded by nobody older than nine." "There'll be parents." "There'll be parents *drinking prosecco out of plastic flutes at two in the afternoon.* By three they'll be crying about their mortgages. That's our window." Bill sat down heavily and stared at the diagram, which Molly had drawn in three colors of highlighter. "Talk me through it." "Fourteen hundred hours, guests arrive. Fourteen twenty, the entertainer shows up. His name is Mister Wobbles." "You've researched Mister Wobbles." "I've *become* Mister Wobbles. I called the agency and cancelled him. Food poisoning. Very sad." Molly produced a red foam nose from her pocket and set it on the table like evidence. "I go in through the front door with a bag of balloons. Nobody questions a clown, Bill. A clown is invisible." "A clown is the opposite of invisible." "A clown is *ignored*, which is better. Meanwhile you're at the side gate in high-vis, checking the gas meter." "There's no gas meter." "Nobody's ever once known where their gas meter is. You say 'gas meter,' they say 'oh, it's round the back somewhere,' and then they go and get you a glass of squash. It's the perfect crime." Bill turned the foam nose over in his fingers. "And the presents?" "Pass-the-parcel." "Pardon?" "I run the games. I insist on pass-the-parcel. I insist it's played with *all* the presents, because"—she raised a finger—"of *inclusivity*. Every child feels seen. The parents will weep. And while the music's going, the parcels go round the circle, out the patio doors, and into your wheelbarrow." "My wheelbarrow." "You'll be doing the flowerbeds by then. You've moved on from the gas meter." Bill was quiet for a long moment. Outside, a bin lorry groaned past. "Molly," he said. "We stole a Vermeer." "We did." "We stole a Vermeer out of a private collection in Antwerp with three seconds to spare and I had to swim a canal." "That was a beautiful night." "And now we're robbing a seven-year-old named Tobias." "Tobias," said Molly, "is the son of Gerald Hames." Bill's chewing stopped. "Gerald Hames," Molly continued, "who bought our Vermeer. Who kept it in a downstairs loo. Who told Interpol our names over a game of squash." She slid a photograph across the table: a small boy in a shirt buttoned to the throat, holding a tennis racket he clearly did not want. "One of those twelve boxes contains a watch worth more than this house. Grandpa's watch. Passed down. Insured for a number I'm not going to say out loud because you'll faint." Bill picked up the red nose and pushed it onto his face. "What size high-vis?" he said.
created_at
"2026-07-25T13:15:55.177Z"
7/25/2026, 1:15:55 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3951
moderation_latency
(null)
generation_time
24919
tokens_prompt
75
tokens_completion
778
native_tokens_prompt
93
native_tokens_completion
1148
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
0.029165
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNqa9wY7G6AWzHxa6DYV"
is_byok
false
latency
3938
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-1784985355-r9Lx5mCvI571iHdMFZU8"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784985355-cx9SlvVHZPF3BQu8NTLt"
upstream_id
"msg_011CdNqa9wY7G6AWzHxa6DYV"
total_cost
0.029165
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
81.27%
Matches word count
n/a
neededClean
false
words
512
0.00%
Dialogue to Total Word Ratio
Ratio: 67.12%, Deviation: 17.12%
neededClean
false
wordsTotal
520
wordsDialogue
349
40.6364%