NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX40GT9K7QJZ5XV4VMRJ0SPB
Run Details
Model:
GPT-5.6 Sol (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
18.3s
Cost:
$0.024840
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
98.71%
Matches word count
(details)
65.20%
Dialogue to Total Word Ratio
(details)
81.9548%
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 a crayon map across the café table. It depicted a house, six stick children, and something either a guard dog or a furious potato. “Bill, explain the potato.” “That’s Mr. Snuffles.” “Mr. Snuffles is spherical?” “He’s a pug. They’re all head.” “And the red scribbles?” “Laser beams.” “A children’s party has laser security?” “No. Those are streamers. But streamers can be surprisingly disorienting.” Bill sipped his tea through a licorice straw. Molly stared until he stopped. “Let’s review,” she said. “At three o’clock, twenty children enter Number Fourteen, Buttercup Lane.” “Correct.” “At three fifteen, the magician arrives.” “Correct.” “At three twenty, he discovers we’ve locked him in a recycling bin.” “Temporarily redirected him.” “In a bin.” “A clean bin.” “You checked?” “I looked at it from across the road.” Molly rubbed her temples. Around them, the café hummed with respectable people committing no crimes whatsoever. “Then you enter dressed as the magician,” she said. “Behold! The Amazing William!” “You cannot call yourself William. Your name is Bill.” “Bill is short for William.” “Exactly. It’s a breadcrumb trail for the police.” “Fine. The Amazing Sebastian.” “You don’t look like a Sebastian.” “What does a Sebastian look like?” “More velvet. Less ear.” Bill tucked his ears beneath his cap. “Meanwhile,” Molly continued, “I enter through the kitchen disguised as the cake decorator.” “Can you decorate cakes?” “I can remove evidence from cakes.” “That means eating them.” “It’s an essential criminal skill.” “And then?” “You distract the children with magic. I locate the presents. We load them into the laundry chute, collect them in the basement, and escape through the garden.” Bill frowned at the map. “There’s no laundry chute.” “You drew one.” “That’s a chimney.” “Why does the chimney go into the basement?” “Architectural whimsy.” Molly turned the map upside down. It did not improve. “How many presents are we talking about?” “Forty, perhaps fifty.” “For one child?” “Apparently little Crispin has wealthy friends.” “Crispin?” “Yes.” “We’re robbing a child named Crispin?” “Don’t get sentimental.” “I’m not sentimental. I’m suspicious. Crispins grow up to own private islands and hunt accountants.” Bill leaned closer. “There is one complication.” “Besides the pug, the nonexistent chute, your fake magic career, and your ears?” “Party bags.” “What about them?” “Every child leaves with one. If we leave without party bags, we’ll stand out.” “We’re adults.” “Exactly. Very tall children without party bags.” Molly slowly folded the map. “New plan. We go in as health inspectors.” “At a birthday party?” “Sugar compliance division.” “Is that real?” “It will sound real if we carry clipboards.” “And the presents?” “Confiscated as dangerously festive.” Bill considered this, nodding gravely. “What about Mr. Snuffles?” “We bribe him.” “With what?” Molly slid a sausage across the table. Bill’s eyes widened. “You’ve thought of everything.” “I always do.” The sausage rolled off the edge. A pug shot from beneath the next table, swallowed it, and vanished. Molly and Bill watched in silence. “Was that him?” Molly asked. Bill checked the crayon map. “Either him or the potato.”
created_at
"2026-07-09T17:58:58.965Z"
7/9/2026, 5:58:58 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2149
moderation_latency
(null)
generation_time
17822
tokens_prompt
77
tokens_completion
1193
native_tokens_prompt
66
native_tokens_completion
817
native_tokens_completion_images
0
native_tokens_reasoning
82
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
"completed"
service_tier
"default"
usage
0.02484
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_01bb1db26c3914b5016a4fe1630bd881979d2492714b0d02ac"
is_byok
false
latency
577
model_permaslug
"openai/gpt-5.6-sol-20260709"
provider_name
"OpenAI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.18.0; linux; x64))"
http_referer
(null)
request_id
"req-1783619938-crNGhwafVa56qDuX2Fnx"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783619938-AmUtmlAnPXKy2Rr7jjoB"
upstream_id
"resp_01bb1db26c3914b5016a4fe1630bd881979d2492714b0d02ac"
total_cost
0.02484
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
data_region
"global"
Evaluation details
Result
Evaluator
Details
Meta Data
98.71%
Matches word count
n/a
neededClean
false
words
506
65.20%
Dialogue to Total Word Ratio
Ratio: 72.92%, Deviation: 2.92%
neededClean
false
wordsTotal
506
wordsDialogue
369
81.9548%