NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37PT2PH4GSZDMJ9RMGYVET3
Run Details
Model:
Claude Opus 5.5 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
23.2s
Cost:
$0.033720
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
93.65%
Matches word count
(details)
0.62%
Dialogue to Total Word Ratio
(details)
47.1365%
Overall
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
Bill spread the map across the hood of the van. It was drawn in crayon, because that was what had been in the glove box. "Right. Target's a seven-year-old named Tyler. Party starts at two. Bouncy castle, magician, forty guests." "Forty guests," Molly said, "means forty presents. Minimum." "Unless some of them go in on a joint gift." "Nobody goes in on a joint gift for a seven-year-old, Bill. That's a grandparent move." "Fair." He tapped a purple rectangle. "Presents go on the gift table, here, by the patio. We need eyes on it the whole time." "Which is why I'm going in as the magician." Bill stared at her. "You can't do magic." "I can make forty presents disappear. That's the whole act." "And when some kid asks you to pull a rabbit out of a hat?" "I tell him the rabbit's in witness protection. Kids respect a code of silence." Bill sighed and rubbed his eyes. "Fine. What am I doing?" "You're delivering the bouncy castle." "They already booked a bouncy castle." "They booked Jumpin' Jerry's Castle Kingdom. We've booked Jumpin' Jerry. He's in the back of the van." Bill glanced over his shoulder. Something in the back of the van thumped twice. "Molly, that's kidnapping." "It's a bribe. I gave him two hundred quid and a packet of Jaffa Cakes. He's having a lovely time." A muffled voice called, "They're the orange ones! Cheers!" Bill lowered his voice. "Okay. Walk me through it." "Two-fifteen, you inflate the castle. Two-thirty, I start the show and every kid in the garden is watching me. Two-forty, you load the presents into the castle." "Into it?" "Through the air valve. It's huge. Then at three, you deflate it, roll it up with the presents inside, and drive off. Nobody questions a man carrying a bouncy castle." "Everybody questions a man carrying a lumpy bouncy castle." "Tell them it's the turrets." Bill frowned at the crayon map for a long moment. "What about the mum?" Molly's face darkened. "Denise." "You said her name like it was a disease." "Denise is chair of the PTA. Denise ran the school fete with military precision. Last year, someone parked in her space at the bake sale and they've never been seen again." "Never?" "Well, they moved to Swindon. Same thing." Bill swallowed. "So how do we handle Denise?" "PiƱata. Three o'clock. She'll be holding the stick. Every parent knows you don't take your eyes off a blindfolded child with a weapon." "That's actually quite clever." "I'm a professional, Bill." He folded the map and hesitated. "One more thing. What are we actually going to do with forty children's presents?" Molly blinked. "Sell them." "To who?" "To..." She paused. "Other children." "Children don't have money, Molly." "Then their parents." "So we're stealing presents from children to sell to the parents of other children, who will give them to children." Molly considered this. "It's a circular economy. Very eco-friendly." From the back of the van, Jerry called, "Can I keep the Lego?" "No!" they said together.
created_at
"2026-09-23T18:00:25.06Z"
9/23/2026, 6:00:25 PM
model
"anthropic/claude-opus-5.5-20260921"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
7029
moderation_latency
(null)
generation_time
23123
tokens_prompt
75
tokens_completion
935
native_tokens_prompt
95
native_tokens_completion
1667
native_tokens_completion_images
(null)
native_tokens_reasoning
536
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
0.03372
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790186425-onhNztVGFITljw7X868x"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790186425-Q78GEfsrX3xqTK9v4jed"
upstream_id
"msg_011CfLooMSu8wZkDPrDgeaju"
provider_responses
0
endpoint_id
"3a2388bc-3740-4e64-a1f5-4e301726b6b9"
id
"msg_011CfLooMSu8wZkDPrDgeaju"
is_byok
false
latency
1839
model_permaslug
"anthropic/claude-opus-5.5-20260921"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.03372
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
93.65%
Matches word count
n/a
neededClean
false
words
509
0.62%
Dialogue to Total Word Ratio
Ratio: 80.08%, Deviation: 10.08%
neededClean
false
wordsTotal
517
wordsDialogue
414
47.1365%