NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-06-01-01KT1HG0PZQJZW47B8Z4FJ56Y7
Run Details
Model:
Claude Opus 4.8 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
21.4s
Cost:
$0.028440
Provider:
Anthropic
Evaluation summary
Score
Evaluator
Details Link
93.65%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
46.8251%
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 blueprint across the kitchen table, smoothing out the creases with the reverence of a surgeon prepping for a transplant. Except the blueprint was a child's drawing, and the surgical site was a backyard. "This," she announced, "is the Hendersons' property. Timmy turns seven on Saturday." Bill squinted at the crayon scrawl. "Why is the house purple?" "Because their daughter drew it. Focus, Bill. The presents will be stacked here, by the bouncy castle." She tapped a green blob that might have been a tree or possibly a dinosaur. Bill leaned back, crossing his arms over his enormous frame, which had once allowed him to wedge through a bank's ventilation shaft and now mostly allowed him to wedge into folding chairs. "Molly, we've robbed casinos. Museums. That one yacht with the gold-plated toilet. And now we're stealing from a seven-year-old?" "Not just a seven-year-old. A *spoiled* seven-year-old. Do you know what these suburban parents buy their kids? Tablets. Drones. One of those hoverboards that catch fire. It's a goldmine wrapped in wrapping paper." The thing about Molly, Bill had learned over fifteen years of partnership, was that she could rationalize anything. Once she'd justified stealing a nun's bicycle by arguing that the exercise would be good for the nun's soul. "And the plan?" he asked, because resistance was futile and dinner was getting cold. Molly produced two crumpled costumes from a duffel bag and tossed one at him. He held it up. It was a clown suit, the fabric a screaming combination of polka dots and existential dread. "Absolutely not." "You'll blend in. Every kid's party has a clown." She was already shrugging into her own outfit—a fairy princess ensemble that strained at the seams. "I'll handle crowd control with balloon animals. You move the merchandise to the van during the piñata distraction. Children go feral around candy. It's nature's smokescreen." Bill examined the clown suit with the deep sorrow of a man confronting his own mortality. "I have a reputation. I was on Interpol's watch list." "You were on Interpol's watch list for forty minutes before they realized they had the wrong Bill." "It still counts." She zipped up her costume and adjusted a tiara that had clearly survived several other heists, judging by the bloodstain on one rhinestone. "The getaway car is parked behind the fence. We grab the gifts during 'Happy Birthday,' which lasts approximately twenty-two seconds—I timed it on YouTube. By the time they cut the cake, we're three counties away, splitting the proceeds and laughing." Bill stood, the floorboards groaning. He pulled the clown suit on over his clothes, the bright red nose dangling from his palm like a tiny, judgmental tomato. "For the record," he said, "if anyone asks, we hit a jewelry store." Molly patted his cheek, leaving a smear of glitter. "Whatever helps you sleep, sweetheart. Now honk your nose. We rehearse the bit at dawn." Bill honked it. The sound was, somehow, the saddest thing he'd ever heard. "Perfect," Molly grinned. "We're going to be *terrific*."
created_at
"2026-06-01T12:10:49.444Z"
6/1/2026, 12:10:49 PM
model
"anthropic/claude-4.8-opus-20260528"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1429
moderation_latency
(null)
generation_time
21331
tokens_prompt
77
tokens_completion
806
native_tokens_prompt
93
native_tokens_completion
1119
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)
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
0.02844
router
(null)
provider_responses
0
endpoint_id
"b0d3beaa-5ff7-4666-a749-b1bb2b14c916"
id
"msg_01LdUuYYzFFChTP9WtBTo1tj"
is_byok
false
latency
1429
model_permaslug
"anthropic/claude-4.8-opus-20260528"
provider_name
"Anthropic"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.16.0; linux; x64))"
http_referer
(null)
request_id
"req-1780315849-2NUihwQGSr4aeaq5Zmfr"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1780315849-bh9MzU1wItlOeK0Iw1Qe"
upstream_id
"msg_01LdUuYYzFFChTP9WtBTo1tj"
total_cost
0.02844
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Anthropic"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
93.65%
Matches word count
n/a
neededClean
false
words
509
0.00%
Dialogue to Total Word Ratio
Ratio: 45.45%, Deviation: 15.45%
neededClean
false
wordsTotal
517
wordsDialogue
235
46.8251%