NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-08-01M4DAM1DDNDRN8NYHR3510H93
Run Details
Model:
Claude Haiku 5.5 (Adaptive)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
10.6s
Cost:
$0.001012
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
3.62%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
1.8117%
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 rain drummed on the roof of the white van parked three houses down from the Pemberton birthday party. Molly squinted through binoculars at the bouncy castle wheezing on the front lawn. "Seven-year-old Tyler's party," she said. "Forty kids, one clown, and a mountain of presents the size of a garden shed." Bill cracked a walnut between his fingers and dropped the shell into the cup holder. "Forty kids is a lot of witnesses, Moll." "Kids don't witness anything. They're too busy eating blue frosting." She lowered the binoculars. "Tell me the plan again." Bill unfolded a blueprint across his knees. It was drawn on the back of a cereal box. "Phase one: we walk in through the side gate dressed as party clowns." "We don't have clown costumes." "We have the clown's car. It's parked in the driveway with the door open." "That's not a plan, Bill. That's a clown car with the door open." He tapped a rectangle labeled GARAGE. "Phase two: the presents are stacked in there. We load them into the van using a ramp." "What ramp?" "The skateboard ramp from my nephew's birthday. I borrowed it." Molly sighed and pulled on a balaclava with a cartoon unicorn stitched across the forehead. "Fine. But if we get caught, I'm telling them it was your idea." "Fair. Oh, and the security system." "Let me guess. A baby monitor." "A baby monitor," Bill confirmed proudly, holding up a plastic gadget the size of a sandwich. "I've rigged it to play lullabies whenever anyone gets too close." Molly stared at him. "That's actually clever." "Thank you." They crept across the lawn at half past three, skirting the bouncy castle, which groaned like an asthmatic dragon. Inside the garage, presents towered in shimmering paper, each tagged with a name in crayon. Molly grabbed a box shaped like a tricycle. Bill hefted a glittering bundle that rattled suspiciously. "Bill," Molly whispered, "what's in that one?" "Says here it's a Lego Death Star. Ages eight and up." "We're robbing a seven-year-old." "We are not robbing a child. We are redistributing his joy to people who need it more." "Who needs it more than a seven-year-old?" "Me. Mostly me. I've never had a Death Star." Somewhere in the house, a door creaked open. Both thieves froze. The baby monitor in Bill's pocket began playing "Twinkle, Twinkle, Little Star" at full volume. "Bill!" "That's not me! That's the monitor!" A tiny voice called from the hallway. "Mommy? Is the clown still here?" Molly grabbed the tricycle box and sprinted for the side gate. Bill scooped up the Death Star, tripped over a squeaky plastic dinosaur, and rolled into a garden hose. By the time they reached the van, the lullaby had switched to a tuba fanfare, and somewhere behind them, a very confused clown was shouting for his car keys.
created_at
"2026-10-08T08:38:32.629Z"
10/8/2026, 8:38:32 AM
model
"anthropic/claude-haiku-5.5-20261007"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
10548
moderation_latency
(null)
generation_time
10548
tokens_prompt
75
tokens_completion
714
native_tokens_prompt
95
native_tokens_completion
2006
native_tokens_completion_images
(null)
native_tokens_reasoning
949
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
0.0010125
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1791448712-bf8zO3aya7rxVRR6Qtpw"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1791448712-uc2wkonTGgTztuTK2HXh"
upstream_id
"msg_011CfpU619WYubWpdwFURfXY"
provider_responses
0
endpoint_id
"b6cdf493-3a61-441f-a134-e7944fddb980"
id
"msg_011CfpU619WYubWpdwFURfXY"
is_byok
false
latency
776
model_permaslug
"anthropic/claude-haiku-5.5-20261007"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.0010125
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
3.62%
Matches word count
n/a
neededClean
false
words
476
0.00%
Dialogue to Total Word Ratio
Ratio: 48.76%, Deviation: 18.76%
neededClean
false
wordsTotal
482
wordsDialogue
235
1.8117%