NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37PCHBD5M3QH7HP6JP50WXF
Run Details
Model:
Claude Opus 5.5 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
42.5s
Cost:
$0.079720
Provider:
Claude Platform on AWS
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
99.46%
Dialogue to Total Word Ratio
(details)
79.8701%
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
Molly spread the blueprints across the hood of the van, pinning one corner with a thermos and the other with Bill's elbow, which he had not volunteered. The blueprints were, strictly speaking, a birthday invitation. Someone had drawn a map on the back in purple crayon, complete with a smiling sun and an arrow labeled PARTY HERE. Molly had circled the arrow three times in red marker, which made it look professional. "Eleven-hundred hours," she said. "The Hendersons' backyard. Thirty-two guests, all under the age of eight. Every single one of them bringing a gift." Bill squinted at the crayon sun. "That's a lot of loot." "That's a mountain of loot, Bill. That's Everest wrapped in dinosaur paper." Bill nodded slowly, the way he did when he was pretending to understand something. He was a large man with a gentle face, the kind of face that made grandmothers hand him their groceries, and he had been Molly's partner for six years. In that time, he had successfully cracked two safes, one of which had been open. "What's the security like?" he asked. Molly produced a surveillance photo. It showed a bouncy castle shaped like a dragon, a folding table covered in juice boxes, and a woman in a sun hat holding a clipboard. "That's Linda Henderson," Molly said. "Mother of the birthday boy. She runs the PTA like a military junta. Nobody gets past Linda." "Nobody at all?" "Last year a man tried to leave a bake sale with an extra brownie. They never found him." Bill swallowed. Molly explained the plan, tapping each point with her marker. Bill would arrive dressed as a professional clown, a role for which he had been rehearsing all week, mostly by falling down. While the children were distracted, Molly would slip in through the side gate, locate the gift table, and transfer everything into the van, which she had painted to read BALLOON DELIVERY in letters that were only slightly crooked. "And then we're rich," she said. "How rich?" Molly did some math on her fingers. It did not go well, so she stopped. "Rich in Lego. Possibly slime," she said firmly. Bill brightened. He had always wanted a Lego pirate ship, but he had never felt comfortable buying one for himself, and nobody else had ever thought to buy him one. It was, he suspected, the tragedy of his life. "What if the kids catch us?" "They're children, Bill. They can't even tie their shoes." "The Henderson boy has a karate belt." Molly paused. She had not known about the karate belt. She looked down at the map, at the smiling crayon sun, which suddenly seemed to be smiling at her rather than with her. "What color?" she asked. "Yellow." "That's fine. Yellow's nothing." "He's six, Molly. A six-year-old with a yellow belt is basically a ninja." Somewhere down the street, a child shrieked with joy, or possibly battle. Molly folded the map very carefully and put it in her pocket. "We'll bring snacks," she said. "Nobody attacks the man with the snacks. That's basic military strategy."
created_at
"2026-09-23T17:53:01.31Z"
9/23/2026, 5:53:01 PM
model
"anthropic/claude-opus-5.5-20260921"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
38178
moderation_latency
(null)
generation_time
42430
tokens_prompt
75
tokens_completion
1370
native_tokens_prompt
95
native_tokens_completion
3967
native_tokens_completion_images
(null)
native_tokens_reasoning
2907
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
0.07972
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790185981-HuF3e3zNeLESyHo6wzk8"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790185981-42ccfeuOlXbnL6sAvquc"
upstream_id
"msg_011CfLoEe6LwTKweUfgm3wgW"
provider_responses
0
endpoint_id
"3a2388bc-3740-4e64-a1f5-4e301726b6b9"
id
"msg_011CfLoEe6LwTKweUfgm3wgW"
is_byok
false
latency
1852
model_permaslug
"anthropic/claude-opus-5.5-20260921"
provider_name
"Claude Platform on AWS"
status
200
total_cost
0.07972
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
99.46%
Dialogue to Total Word Ratio
Ratio: 29.67%, Deviation: 0.33%
neededClean
false
wordsTotal
519
wordsDialogue
154
79.8701%