NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37Q0Z1PKTQV32Q9Q6B5HN65
Run Details
Model:
Claude Opus 5.5 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
30.6s
Cost:
$0.049260
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
0.08%
Matches word count
(details)
0.73%
Dialogue to Total Word Ratio
(details)
0.4089%
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
Molly unrolled the blueprint across the kitchen table and pinned the corners with four mismatched mugs. "Right. The Hendersons' house. Saturday, two o'clock. Tyler turns seven." "Happy birthday, Tyler," said Bill solemnly. "Don't get attached. By four o'clock, Tyler's going to learn that life is cruel and presents are temporary." "That's a bit dark, Molly." "It's character-building. We're basically educators." She tapped the living room. "Gift table's here, by the window. Forty kids, forty presents, minimum. Posh neighbourhood. I'm talking drones, tablets, those robot dogs that cost more than my car." "Your car cost two hundred quid." "Which is why I need a robot dog, Bill." Bill squinted at the map. "What's this square in the garden?" "Bouncy castle. That's our distraction. At exactly two forty-five, you let the air out of it." Bill gasped. "With the kids inside?" "They'll be fine. It's like being hugged by a very large, very sad balloon. Every parent runs outside, and I stroll out the front door with the gifts." "And how do we get in? They won't just let two grown adults wander in off the street." Molly smiled and dumped a plastic bag onto the table: a rainbow wig, a red nose, and a pair of shoes roughly the size of canoes. "No," said Bill. "Yes." "Molly, I'm scared of clowns." "You won't see the clown. You'll be the clown." "That's worse! I'll be scared of my own reflection!" "Then avoid mirrors. You're Bingo the Birthday Buffoon. Balloon animals, silly jokes, the works." "I can't make balloon animals." "Everyone can make a snake, Bill. You blow up a balloon and don't do anything to it." He considered this. "What are you going as?" "The magician. Top hat, cape, big box. The Great Mollini." "Why do you get to be the magician?" "Because the magician has a box. And what fits in a box?" Bill thought hard. "A rabbit?" "Presents, Bill. I make them disappear. Literally. Greatest magic trick in history." "What if the kids want them to reappear?" "Then I tell them the magic was inside their hearts all along." Bill fiddled with the red nose. "There's just one problem." "There are no problems. This plan is airtight." "Unlike the bouncy castle." "Exactly. What's the problem?" "The Hendersons already hired a clown. I saw the booking on their fridge when I was scouting. Some bloke called Chuckles." Molly went very still. "How did you see their fridge?" "I said I was reading the meter." "Which meter?" "The fridge meter." She closed her eyes and breathed slowly through her nose. "Fine. We deal with Chuckles." "Deal with him how?" "Bill, we're going to have to do the one thing no professional thief ever wants to do." He swallowed. "What?" Molly picked up the phone with grim determination. "Book him for a different party."
created_at
"2026-09-23T18:04:10.696Z"
9/23/2026, 6:04:10 PM
model
"anthropic/claude-opus-5.5-20260921"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
7387
moderation_latency
(null)
generation_time
30480
tokens_prompt
75
tokens_completion
993
native_tokens_prompt
95
native_tokens_completion
2444
native_tokens_completion_images
(null)
native_tokens_reasoning
1386
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
0.04926
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790186650-tg9A2DZf1aCmVkSC1iiT"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790186650-sNdMeT015vUu2bSsaiRs"
upstream_id
"msg_011CfLp5zACxkocBeK6i5ukP"
provider_responses
0
endpoint_id
"3a2388bc-3740-4e64-a1f5-4e301726b6b9"
id
"msg_011CfLp5zACxkocBeK6i5ukP"
is_byok
false
latency
1883
model_permaslug
"anthropic/claude-opus-5.5-20260921"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.04926
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
0.08%
Matches word count
n/a
neededClean
false
words
471
0.73%
Dialogue to Total Word Ratio
Ratio: 79.92%, Deviation: 9.92%
neededClean
false
wordsTotal
473
wordsDialogue
378
0.4089%