NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-05-31-01KSYXPY1QRE2T7KC061WTBGVG
Run Details
Model:
Claude Opus 4.8 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
21.7s
Cost:
$0.029490
Provider:
Anthropic
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
20.40%
Dialogue to Total Word Ratio
(details)
60.1595%
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 spread the blueprints across the kitchen table, smoothing the wrinkled paper with both hands. "Okay, here's the layout," she said. "Bouncy castle here, snack table here, and the presents—" she stabbed a finger at the corner, "—right here, by the window." Bill leaned over, squinting. "Why is there a drawing of a clown?" "That's the clown. Pedro the Magnificent. He's the wildcard." "A clown is the wildcard?" "Bill, never underestimate a man who can make balloon animals. Those hands are fast." Bill straightened up, crossing his arms. "I still don't love this. We're professionals. We robbed a casino last year. Now we're stealing from a seven-year-old?" "His name is Tyler, and he's turning eight," Molly corrected. "And those presents are practically gift-wrapped cash. Premium toys, sealed in boxes, completely untraceable." "We're going to fence Lego sets?" "The market is booming, Bill. Do you know what a vintage Millennium Falcon goes for?" He paused. "...No." "Eight hundred dollars. And little Tyler's got rich relatives." Bill rubbed his chin, considering. "Alright. What's our cover?" "We blend in. You're the entertainer, I'm the catering." "I am not dressing as a clown." "You're not the clown," Molly sighed. "You're the magician. You wear a cape. It's dignified." "A cape is dignified?" "More dignified than juggling, yes." Bill threw up his hands. "And what do I actually do? I don't know any magic!" "You don't need to know magic. You need to distract twelve sugar-fueled children for ninety seconds while I load the presents into the catering van." "Ninety seconds is a long time with children, Molly." "That's why you've got the doves." Bill blinked. "What doves?" "The doves in the box behind you." Bill turned slowly and stared at a cardboard box that was, indeed, gently cooing. "Why are there doves in my kitchen?" "They were on sale." "Doves were on sale?" "Everything's negotiable if you don't ask questions." Molly tapped the blueprint again. "Now focus. At exactly two-fifteen, the parents gather for cake. That's our window. Kids are seated, adults are distracted, and Pedro takes his smoke break out back." "The clown smokes?" "Pedro contains multitudes. The point is, the room's unguarded for one beautiful minute." Bill leaned in, finally interested. "And the getaway?" "Catering van out front. We've got balloons on the side that say 'Party Time Catering.' Nobody questions a van with balloons." "That's actually clever." "I have my moments." She rolled up the blueprints. "So we're agreed? Two-fifteen, you do the doves, I grab the loot, we vanish before the candles are blown out." Bill exhaled slowly, then nodded. "Fine. But I have one condition." "Name it." "If a single child cries, we abort." Molly stared at him. "You're going soft." "I have a heart, Molly." "You stole a kidney once." "That was business," Bill said. "This is principle." She held out her hand. "Deal. No tears, no problem." He shook it. Behind them, the doves cooed ominously, as if they already knew the whole thing would go terribly, terribly wrong.
created_at
"2026-05-31T11:46:35.715Z"
5/31/2026, 11:46:35 AM
model
"anthropic/claude-4.8-opus-20260528"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1118
moderation_latency
(null)
generation_time
21671
tokens_prompt
77
tokens_completion
842
native_tokens_prompt
93
native_tokens_completion
1161
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
0.02949
router
(null)
provider_responses
0
endpoint_id
"b0d3beaa-5ff7-4666-a749-b1bb2b14c916"
id
"msg_01FfFn4aWQAYo94pFvzPVPRY"
is_byok
false
latency
1118
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-1780227995-oAtaraTY8DmVFw5l8k8y"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1780227995-NAdAQhDue6scovHKNFz6"
upstream_id
"msg_01FfFn4aWQAYo94pFvzPVPRY"
total_cost
0.02949
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Anthropic"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
99.92%
Matches word count
n/a
neededClean
false
words
503
20.40%
Dialogue to Total Word Ratio
Ratio: 75.64%, Deviation: 5.64%
neededClean
false
wordsTotal
509
wordsDialogue
385
60.1595%