NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37PR9G6XXZMQRG6NNAT23E0
Run Details
Model:
Claude Opus 5.5 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
33.9s
Cost:
$0.056340
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
98.95%
Dialogue to Total Word Ratio
(details)
79.6119%
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 van smelled of gas-station burritos and bad decisions, which, as far as Molly was concerned, meant it smelled like home. She unrolled the blueprint across the dashboard. It was not, strictly speaking, a blueprint. It was a birthday invitation with a dinosaur on it, and someone had drawn arrows all over the dinosaur in purple crayon. Bill leaned over from the driver's seat. He was a large man with a small mustache, and both of them looked worried. "Is that the layout?" he asked. "It's the invitation," Molly said. "Tyler is turning six. Party starts at two. Presents go on the gift table by the bouncy castle. That's our target." "How many presents are we talking?" "Twenty-two kids confirmed. Parents in this neighborhood? Minimum two gifts each. We're looking at forty-four boxes of pure, unsupervised plastic." Bill whistled low. Then he frowned, which on Bill took a while, like watching a cloud decide to rain. "What's the entry point?" Molly tapped a crayon arrow. "Side gate. Latch is broken. I checked on Tuesday when I pretended to be selling kitchen knives." "You sold them knives?" "I sold them three knives. That's not the point." Outside, a minivan the color of a boiled shrimp pulled into the cul-de-sac. A woman climbed out hugging a present the size of a refrigerator. Bill watched it pass with the expression of a man seeing the face of God. The plan was elegant in its simplicity. Bill would enter wearing a rented clown suit. Nobody questions a clown at a children's party; clowns are like weather, simply something that happens to you. While the children were distracted, Bill would carry the presents through the side gate and into the van. The weakness of the plan was Bill. "I've been practicing my clown stuff," he said proudly, pulling a limp balloon from his pocket. He blew until his face matched the shrimp van, then twisted it with tremendous concentration. It popped. "That was a giraffe," he said. "It was a tragedy," Molly said. "Just juggle." "I can't juggle." "Then fall down. Kids love it when clowns fall down." Bill considered this and found it within his skill set. At two o'clock precisely, the bouncy castle inflated with a sound like a great beast waking. Bill wobbled out in oversized shoes, honked his nose twice for courage, and started across the lawn. He made it four steps before a pack of six-year-olds spotted him and charged, shrieking with joy. Through her binoculars, Molly watched her partner vanish beneath a wave of tiny bodies and party hats. Forty minutes later, the side gate creaked open. Bill staggered out, face paint smeared, holding nothing but a paper plate of cake. "Where are the presents?" Molly demanded. "They made me help open them," he whispered. "Tyler got a drone. It's very good. He let me fly it." "Bill." "Also, his mom booked me for Saturday. Two hundred bucks, plus cake." Molly stared at him for a long moment, then at the cake. "What kind of cake?" "Chocolate." She sighed and reached for a fork.
created_at
"2026-09-23T17:59:26.486Z"
9/23/2026, 5:59:26 PM
model
"anthropic/claude-opus-5.5-20260921"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
5952
moderation_latency
(null)
generation_time
33882
tokens_prompt
75
tokens_completion
1185
native_tokens_prompt
95
native_tokens_completion
2798
native_tokens_completion_images
(null)
native_tokens_reasoning
1711
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
0.05634
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790186366-A5FBPVs13nXjW50TerKe"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790186366-dNoqrgjP7QZTgxURe1gT"
upstream_id
"msg_011CfLoj2mdYQu9LZZCPAN9s"
provider_responses
0
endpoint_id
"3a2388bc-3740-4e64-a1f5-4e301726b6b9"
id
"msg_011CfLoj2mdYQu9LZZCPAN9s"
is_byok
false
latency
1777
model_permaslug
"anthropic/claude-opus-5.5-20260921"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.05634
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Claude Platform on AWS"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
60.28%
Matches word count
n/a
neededClean
false
words
515
98.95%
Dialogue to Total Word Ratio
Ratio: 30.46%, Deviation: 0.46%
neededClean
false
wordsTotal
522
wordsDialogue
159
79.6119%