NC Bench
Overview
About
Tests
Models
Model
Claude Haiku 4.5
Test
Dialogue tags
Scenario
Write 500 words with 70% dialogue
Run ID
2026-02-08-01KGYZ75DSY3JBEHPNEW3JT184
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 adjusted her fake mustache as Bill spread out the blueprint across the kitchen table. "So you're telling me," Bill said, "that we're robbing a seven-year-old's birthday party? We've hit museums, casinos, and a minor league baseball stadium, and now we're going after a suburban home with juice boxes?" "Not just any suburban home," Molly corrected, jabbing her finger at the diagram. "The Henderson residence. Three kids' birthday parties are happening simultaneously. We're talking at least forty presents, Bill. Forty. Do you know what that's worth?" "Emotional trauma to children?" Bill suggested, pulling out a sandwich from his pocket. "I'm not eating this, by the way. I'm just extremely anxious about what we're doing here." "You were anxious about the jewelry heist too, and you wore those diamond earrings to the grocery store for a week afterward." Molly leaned back in her chair. "Look, entry point is through the back patio. The bouncy castle will provide cover." "The bouncy castle," Bill repeated slowly. "Our cover is an inflatable castle. We've officially hit rock bottom. What happened to us, Molly? We had style. We had panache. We once stole a Rembrandt." "We needed money," Molly said flatly. "Besides, a Rembrandt doesn't pay like it used to. But do you know how much those gaming consoles cost now? Ridiculous." She tapped the blueprint again. "I'll go in as the face painter. You'll be the balloon animal guy." "I don't know how to make balloon animals," Bill protested, biting into his anxiety sandwich anyway. "Neither do I, that's what makes it perfect. They'll be so confused they won't even notice us loading the presents into the van." Molly pulled out two nametags. "You're Steve. I'm Deborah." "Deborah? Really? I'm Steve? We sound like insurance adjusters," Bill moaned. "What about our exit strategy?" "The piƱata is filled with confetti, not candy. I filled it with confetti. When it breaks, everyone will be distracted by the mess and the existential disappointment of the children." Bill set down his sandwich carefully. "That's genuinely dark, Molly. I'm impressed despite myself." He picked up the Steve nametag and stared at it. "If we get caught, I'm telling the police this was all your idea." "Obviously," Molly said, already putting on her Deborah nametag. "I'll tell them you threatened me with balloon animals. Very believable." "This is what we've become," Bill sighed, standing up. "Two professional criminals arguing about who takes the fall for stealing Hot Wheels at a children's party." He grabbed his balloon kit. "What time do we hit the place?" "Saturday, two o'clock sharp. We're going to be the worst birthday entertainers they've ever hired, but we're going to be rich doing it." Bill nodded grimly and bit into his sandwich again.
created_at
"2026-02-08T15:51:47.678277+00:00"
Sunday, February 8th, 2026 at 3:51:47 PM GMT+00:00
model
"anthropic/claude-4.5-haiku-20251001"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
730
moderation_latency
177
generation_time
7487
tokens_prompt
61
tokens_completion
626
native_tokens_prompt
75
native_tokens_completion
704
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)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.003595
router
(null)
provider_responses
0
id
"0127c703-d4a8-41ea-9010-008a3228ee1b"
status
200
is_byok
false
latency
730
endpoint_id
"d000261c-5aeb-47d1-ac87-ad64dbe1cdb6"
provider_name
"Amazon Bedrock"
model_permaslug
"anthropic/claude-4.5-haiku-20251001"
api_type
"completions"
id
"gen-1770565899-v2tnGkUB4yvNN3NVKv0h"
upstream_id
"0127c703-d4a8-41ea-9010-008a3228ee1b"
total_cost
0.003595
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
Evaluation details
Result
Evaluator
Details
Meta Data
0.0000%
Matches word count
n/a
neededClean
false
words
458
12.7414%
Dialogue to Total Word Ratio
Ratio: 76.42%, Deviation: 6.42%
neededClean
false
wordsTotal
458
wordsDialogue
350
6.3707%