NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-25-01KYCRE0V3DZ01SM58JNHZSTJC
Run Details
Model:
Claude Opus 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
22.4s
Cost:
$0.028690
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
2.01%
Matches word count
(details)
98.17%
Dialogue to Total Word Ratio
(details)
50.0890%
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, technically, a place mat Bill had traced with a crayon. He had labeled the bouncy castle THE FORTRESS and drawn an arrow to a lopsided rectangle marked LOOT ZONE, which was actually the folding table where Mrs. Pemberton would arrange the gifts in ascending order of size, because Mrs. Pemberton was that sort of woman. Molly circled the table twice, the way a shark circles a paddling pool, and tapped the crayon rectangle with one chewed fingernail. "Twelve presents," she said. "Fourteen if the Havisham twins come, and they always come, because their mother has no shame." "Fourteen," Bill repeated reverently. "Now tell me the plan." "I go in as the clown." "You go in as the clown." "And you go in as the balloon lady." The kettle shrieked. Molly ignored it, which was a bad sign, because Molly ignored very little. She sat down, folded her hands, and looked at Bill with the terrible patience of a woman who had once shoplifted a chandelier. The problem, as she saw it, was not the party. The party was a soft target: sixteen children under the age of eight, three adults, and a magician named Colin who was known throughout the borough for being drunk by half past two. The problem was Bill, who had, at their last job, hidden in a wardrobe for ninety minutes before discovering it was a shower cubicle. "The clown's already booked," she said. "His name is Bongo. He has a van." "Then I'll be a second clown." "There's never a second clown, Bill. Nobody has ever, in the history of birthdays, hired two clowns." "Rich people might." She considered this. The Pembertons had a water feature in the front garden shaped like a swan being strangled by a cherub. It was not impossible. "Fine. You're the second clown. Bongo won't argue, he's terrified of confrontation, I read his reviews. You keep the parents busy with something long and boring—do the one where you fold the sausage dog wrong and cry." "That's not a bit." "It's a bit now." Molly's own role was simpler. She would arrive at 2:15 with a bin bag, wearing the high-visibility vest she had stolen from a man laying cables, because nothing on earth commands more authority at a children's party than a woman with a bin bag and a lanyard. She would announce, loudly, that the presents required "sorting for allergens." She had tested this line on four separate mothers at the school gates and not one had asked what it meant. "And the gate?" Bill said. "Propped. Bongo's van doors." "And if the birthday boy sees us?" Molly poured the tea at last, sliding a mug across the crayon blueprint, right over the bouncy castle. "Then we tell him the presents have gone to a farm," she said, "where they'll be very happy." Bill nodded slowly, the way a man nods when he has decided something is fine. Outside, somewhere across the estate, a distant ice cream van played eight bars of a song and then stopped, as if it too were casing the neighborhood.
created_at
"2026-07-25T13:46:30.123Z"
7/25/2026, 1:46:30 PM
model
"anthropic/claude-opus-5-20260723"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2990
moderation_latency
(null)
generation_time
22300
tokens_prompt
75
tokens_completion
817
native_tokens_prompt
93
native_tokens_completion
1129
native_tokens_completion_images
(null)
native_tokens_reasoning
58
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
0.02869
router
(null)
provider_responses
0
endpoint_id
"76cb4608-f48c-483d-8da8-9957fb44244e"
id
"msg_011CdNsuP9HvjRctKDxHQp5J"
is_byok
false
latency
1227
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-1784987190-HFDyNb24Xg8BsihOlG89"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1784987190-d7xMMZsU2JUGcKvM9kMr"
upstream_id
"msg_011CdNsuP9HvjRctKDxHQp5J"
total_cost
0.02869
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
2.01%
Matches word count
n/a
neededClean
false
words
525
98.17%
Dialogue to Total Word Ratio
Ratio: 30.61%, Deviation: 0.61%
neededClean
false
wordsTotal
526
wordsDialogue
161
50.0890%