NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX40XN370DB8CA2PWYYSD4QF
Run Details
Model:
GPT-5.6 Sol (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
57.9s
Cost:
$0.151890
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
99.80%
Dialogue to Total Word Ratio
(details)
99.9017%
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 stolen floor plan across Bill's knees while their van idled opposite the community hall, wheezing exhaust like an embarrassed dragon in winter. “The target is six years old, wears a paper crown, and controls thirty-seven presents. Security consists of parents, balloons, and one suspicious grandmother named Doris.” Bill studied the plan upside down, nodding with professional gravity. A purple balaclava covered his face, though he had forgotten to cut eyeholes entirely again. “I dislike grandmothers. They notice everything. Mine identified me during that bank job from my posture, getaway bicycle, and sequined jacket alone on television immediately.” Molly turned the plan around. Their entrance, marked with a skull, was actually the kitchen. Their exit, marked escape, appeared to be a broom cupboard. “We enter disguised as entertainers. You juggle, I perform magic, then we load gifts into the cake trolley during the birthday song finale and vanish.” Bill pulled off the balaclava, revealing face paint, one false eyebrow, and a red nose that squeaked whenever he inhaled through his left nostril loudly. “Small problem: I cannot juggle. Last time I tried, three oranges hospitalized a vicar, and the fourth is probably still orbiting Birmingham somewhere even now.” Across the street, children stampeded into the hall wearing pirate hats. Each carried a present, except one boy dragging an enormous stuffed crocodile behind him. “Forget juggling. Hold this wand, shout mysterious nonsense, and look confident. Children accept any absurdity if an adult says it beside glitter and dry ice.” Molly opened a suitcase containing smoke bombs, lock picks, rabbit ears, and thirty meters of bunting. Bill lifted the wand. Its silver tip flashed ominously. “Why did it flash? Wands should not flash unless they are magical, explosive, or manufactured by that discount spy shop near the old sewage works.” A jet of sparks shot through the van roof. Outside, the stuffed crocodile toppled. Its owner stared at the smoking hole, then began applauding enthusiastically. “Excellent. We have our act. You are Blazing Bill, wizard of insurance claims. I am Madame Molly, mistress of strategic cake relocation and parcel disappearance.” They crossed beneath a banner reading HAPPY BIRTHDAY, ALFIE. Molly pushed the cake trolley. Bill followed, wand hidden inside a limp balloon giraffe for safety. “Remember, take only wrapped presents. Unwrapped toys are evidence. Also, avoid anything that barks, leaks, ticks, or requests batteries in seventeen languages, including Welsh twice.” At the door stood Doris, compact and silver-haired, holding a clipboard like an axe. Her stare pinned them. Molly whispered to Bill, who swallowed hard. “Password?” “You said there was no password.” “There is always a password when someone owns a clipboard.” “Then say something educational, wholesome, and quickly convincing.” Bill straightened, squeaking his nose. Behind Doris, mountains of bright parcels gleamed beside the cake. He raised the wand, which began fizzing softly but alarmingly. “We represent the Royal Academy of Advanced Birthday Mathematics. We teach subtraction. For instance, thirty-seven presents minus thirty-seven presents equals a valuable lesson in sharing.”
created_at
"2026-07-09T18:05:59.285Z"
7/9/2026, 6:05:59 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
49649
moderation_latency
(null)
generation_time
57274
tokens_prompt
77
tokens_completion
1162
native_tokens_prompt
66
native_tokens_completion
5052
native_tokens_completion_images
0
native_tokens_reasoning
4387
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
0.15189
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_05d34bc20f8f6ea5016a4fe307579c8196b306655a41c2dbef"
is_byok
false
latency
237
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-1783620359-sTrvkmGUqOVyqGxdpkXM"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783620359-H28NQrFwRpSqVTCteGBi"
upstream_id
"resp_05d34bc20f8f6ea5016a4fe307579c8196b306655a41c2dbef"
total_cost
0.15189
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
100.00%
Matches word count
n/a
neededClean
false
words
500
99.80%
Dialogue to Total Word Ratio
Ratio: 50.20%, Deviation: 0.20%
neededClean
false
wordsTotal
504
wordsDialogue
253
99.9017%