NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-01-01KWEAB18TYT6Y7FASGTANRZW4
Run Details
Model:
Claude Sonnet 5 (Reasoning, Low)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
18.7s
Cost:
$0.011656
Provider:
Amazon Bedrock
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
49.9595%
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
The warehouse smelled like old cardboard and Bill's cologne, which Molly had repeatedly told him smelled like a gas station bathroom trying to apologize for something. She spread the blueprints across a crate, using a half-eaten donut as a paperweight. "So the target is the Hendricks house," she said. "Timmy Hendricks. Turning seven. Word on the street is he's getting a PlayStation, a bike, and—get this—an actual pony." Bill leaned over the map, squinting like it owed him money. "A pony. Who gets a pony for their birthday? What is he, a Kardashian?" "Apparently his dad owns a mattress store." "Ah. Mattress money. The dirtiest money there is." Bill traced a finger along the blueprint, which was actually just a printout from Google Maps with "BAD GUY ENTRANCE" scrawled in red marker over the garage. "Okay, so we go in through here, past the bounce house—" "There's a bounce house?" "There's always a bounce house, Molly. Rich kids have bounce houses like the rest of us have regret." Molly rubbed her temples. This was their eleventh heist together, and somehow Bill's plans always devolved into personal grievances about children's entertainment. "Focus. We need the guest list, the present table location, and an exit strategy that doesn't involve you eating cake at the crime scene like last time." "That cake was evidence. I was disposing of evidence." "You had frosting on your ski mask, Bill. In the getaway car." He had the decency to look mildly ashamed, which for Bill meant staring at his shoes for exactly two seconds before continuing. "Fine. New rule: no snacking during active heists." "Thank you." "Unless it's really good cake." "Bill." "I'm just saying, if it's like a red velvet situation, we might need to renegotiate terms." Molly decided to let it go, because arguing with Bill about cake ethics was a conversation with no exit ramp. She pointed to a section of the map labeled, in Bill's handwriting, "CLOWN ZONE - PROCEED WITH CAUTION." "What's this?" "There's a clown. Danny the Delightful, apparently. He does balloon animals." "So we avoid the clown." "We can't avoid the clown, Molly, he's clearly the muscle. Look at the way he's positioned near the present table. That's not decorative. That's strategic." "He's a children's entertainer." "That's exactly what he wants you to think." Molly stared at him for a long moment, wondering, as she often did, why she continued to partner with a man who considered a balloon artist a legitimate security threat. But then she remembered the mortgage payment sitting untouched on her kitchen counter, and the fact that stolen presents resold surprisingly well on the internet to desperate parents who'd forgotten birthdays entirely. "Fine," she said. "We neutralize the clown. Diplomatically." "I brought extra balloons. In case we need to blend in." "You brought balloons to a heist." "I brought *options*, Molly. That's the difference between us and amateurs." She looked at the blueprint, then at Bill's hopeful, frosting-stained face, and wondered exactly which category she belonged to.
created_at
"2026-07-01T07:47:17.412Z"
7/1/2026, 7:47:17 AM
model
"anthropic/claude-sonnet-5-20260630"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3985
moderation_latency
(null)
generation_time
18667
tokens_prompt
77
tokens_completion
862
native_tokens_prompt
93
native_tokens_completion
1147
native_tokens_completion_images
(null)
native_tokens_reasoning
45
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
0.011656
router
(null)
provider_responses
0
endpoint_id
"37b16f45-eeae-45f3-9b85-2137f7c6d61b"
id
"msg_01N1sJg5M1pCHo2qmvz4bMrW"
is_byok
false
latency
2513
model_permaslug
"anthropic/claude-sonnet-5-20260630"
provider_name
"Amazon Bedrock"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.17.0; linux; x64))"
http_referer
(null)
request_id
"req-1782892037-24amuc4ndg92OcKhiVPL"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1782892037-Q3vM2Tq7Izafnizz9MbI"
upstream_id
"msg_01N1sJg5M1pCHo2qmvz4bMrW"
total_cost
0.011656
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Amazon Bedrock"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
99.92%
Matches word count
n/a
neededClean
false
words
503
0.00%
Dialogue to Total Word Ratio
Ratio: 52.87%, Deviation: 22.87%
neededClean
false
wordsTotal
505
wordsDialogue
267
49.9595%