NC Bench
Overview
About
Tests
Models
Vendors
Compare
Run: 2026-02-24-01KJ8VEFTKWEDFSZYH6GVC54PB
Run Details
Model:
Claude Sonnet 4.6 (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
27.8s
Cost:
$0.027741
Provider:
Anthropic
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
0.73%
Dialogue to Total Word Ratio
(details)
30.5031%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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 hand-drawn map across the hood of the van and smoothed out the creases with her palm. The map was covered in crayon drawings of balloons and what appeared to be a poorly rendered clown. "Where did you even get this?" Bill asked, squinting at it. "I drew it from memory," Molly said proudly. "I attended the birthday party last year as a guest." "You went to a seven-year-old's birthday party." "I was gathering intelligence, Bill. That's what professionals do." Bill picked up the map and tilted it sideways. "Why is everything purple?" "It's the only crayon I had," Molly said, snatching it back. "Now focus. The presents are going to be piled in the living room, right next to the window. Standard configuration for a suburban birthday party attended by children who have too many things." Bill leaned against the van and folded his arms. "And the security situation?" "One golden retriever named Biscuit, possibly two grandmothers, and a teenage cousin on his phone who won't look up for anything short of a nuclear event." She tapped the map. "We go in through the side gate at fourteen hundred hours, which is when they cut the cake." "Why during the cake?" "Because Bill, no child on this earth looks away from birthday cake. We'll have a four-minute window minimum. Possibly five if there are trick candles." Bill was quiet for a moment. A balloon drifted lazily across the street from somewhere nearby, and they both watched it go. "I want to go on record," Bill said, "as saying this is the strangest job we've ever taken." "Stranger than the cruise ship casino?" "At least the cruise ship casino didn't have a bouncy castle as a potential obstacle." Molly clicked her pen and made a note on the map. "The bouncy castle is actually an asset. The children will be gravitationally committed to it. Motion, energy, chaos. Nobody is watching the presents." "What about the clown?" Bill asked. Molly's expression shifted almost imperceptibly. "We don't discuss the clown." "Molly." "He's a wildcard. We account for wildcards by moving fast." She rolled up the map with crisp efficiency. "In and out in under four minutes. We take the presents, we leave a fruit basket as a goodwill gesture, and we're gone before anyone notices." "A fruit basket," Bill repeated flatly. "I'm not a monster, Bill." He rubbed his face with both hands and looked at the sky like he was hoping for some kind of divine intervention. None arrived. The distant sound of children shrieking with delight floated over the neighborhood fence, carried on a warm Saturday breeze. "Fine," he said finally. "But I'm driving the van." "You always drive the van." "And yet it never feels like enough control," Bill muttered, climbing in. Molly tucked the map into her jacket and smiled. It was a good plan. It was absolutely a good plan.
created_at
"2026-02-24T22:13:51.577Z"
2/24/2026, 10:13:51 PM
model
"anthropic/claude-4.6-sonnet-20260217"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
615
moderation_latency
(null)
generation_time
27734
tokens_prompt
77
tokens_completion
1864
native_tokens_prompt
92
native_tokens_completion
1831
native_tokens_completion_images
(null)
native_tokens_reasoning
1112
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.027741
router
(null)
provider_responses
0
endpoint_id
"6d408764-7dd7-4626-bb87-a6cc1589bc86"
id
"msg_011TXHKRTyef1AqN93tvYVpz"
is_byok
false
latency
598
model_permaslug
"anthropic/claude-4.6-sonnet-20260217"
provider_name
"Anthropic"
status
200
api_type
"completions"
id
"gen-1771971231-ERDHu5mXF7xKCMpqNl9z"
upstream_id
"msg_011TXHKRTyef1AqN93tvYVpz"
total_cost
0.027741
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Anthropic"
Evaluation details
Result
Evaluator
Details
Meta Data
60.28%
Matches word count
n/a
neededClean
false
words
485
0.73%
Dialogue to Total Word Ratio
Ratio: 59.92%, Deviation: 9.92%
neededClean
false
wordsTotal
489
wordsDialogue
293
30.5031%