NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX41F42XA6WWNSHPASFZEFR8
Run Details
Model:
GPT-5.6 Sol (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
1m 38s
Cost:
$0.200160
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
98.47%
Dialogue to Total Word Ratio
(details)
99.2343%
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 crayon map across the bonnet of their borrowed ice-cream van. Beneath the community-hall windows, dragons fluttered, and thirty sugar-fueled children screamed with the intensity of tiny opera singers. “Right,” she said, “the presents are behind the cake table. We enter through the kitchen, cross beneath the balloon arch, and leave before anyone notices two adults crawling under furniture.” Bill peered at the map. “Why is the kitchen drawn as a volcano?” “Because I ran out of gray.” “You had brown.” “Brown is for danger.” “And the enormous purple duck?” “Also danger.” Bill adjusted his paper hat, folded from a ticket, and frowned. “I look like a bankrupt admiral serving soup in a prison during a thunderstorm.” “We need disguises. I’m the magician.” “You don’t know magic.” “I can remove a watch without touching the clasp.” “That’s pickpocketing.” “With confidence, it’s magic. You’re the caterer.” “I don’t cook.” “You don’t need to. Carry this tray of sausage rolls and look disappointed.” “I always look disappointed.” “Exactly. Natural talent.” Bill lifted the foil. “These are rubber.” “Real sausage rolls get eaten.” “So will these.” “Then the children learn an important lesson about chewing.” A burst of applause shook the hall. A rabbit-costumed man staggered past the glass beneath a tower of pink boxes. “Behold the rabbit, Bill. He carries the premium loot in pink boxes.” “There,” Molly whispered. “The rabbit has the premium loot.” “How can you tell?” “Ribbon density.” “Could be party hats.” “Party hats don’t require an armed escort.” “That woman has a serving spoon.” “She knows how to use it. We stick to the plan.” “Which plan? The volcano plan or the duck plan?” “The volcano gets us in. The duck gets us out.” “I’m beginning to suspect you made this up.” “I’m beginning to suspect your hat isn’t regulation catering wear.” Bill frowned, then noticed a small face pressed to the hall window. A girl stared directly at them, holding a plastic sword and wearing the expression of an auditor. “We’ve been spotted.” “Smile.” “You smile.” “My smile is wanted in three counties.” Bill attempted a grin. The girl raised her sword. “She’s calling reinforcements.” “She’s six.” “Six-year-olds travel in packs. Start the engine.” “We haven’t stolen anything.” “We’ve stolen suspicion. It’s surprisingly heavy.” “Wait. She’s pointing at the side door.” “And now she’s miming a key.” “She wants us to break in.” “Or rescue her from the magician.” “You’re the magician.” “Then her judgment is excellent.” A side door opened. The rabbit dropped a pink box. “Fate has delivered a box. Technically, we’re rescuing it from weather and poor rabbit management practices.” “Success,” she said. “We haven’t entered.” “Best kind of burglary.” Molly opened the box. “Inside are twenty identical party whistles.” Bill picked one up. “Premium loot?” “Limited edition.” “They have ducks on them.” Molly looked at the map, then whistle, then hall. Thirty children appeared at the window. “The duck was danger.” “Drive,” said Bill. “Finally, something on this map makes sense.”
created_at
"2026-07-09T18:15:31.686Z"
7/9/2026, 6:15:31 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
17476
moderation_latency
(null)
generation_time
97714
tokens_prompt
77
tokens_completion
1242
native_tokens_prompt
66
native_tokens_completion
6661
native_tokens_completion_images
0
native_tokens_reasoning
5962
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
0.20016
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_0caa0e43c5c3987e016a4fe5458d18819586825b306dd8dc27"
is_byok
false
latency
3803
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-1783620931-Ho6dTE2aUOZgZJUbqNcN"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783620931-bA29uc1U7lLi1u2Xm71b"
upstream_id
"resp_0caa0e43c5c3987e016a4fe5458d18819586825b306dd8dc27"
total_cost
0.20016
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
98.47%
Dialogue to Total Word Ratio
Ratio: 69.44%, Deviation: 0.56%
neededClean
false
wordsTotal
504
wordsDialogue
350
99.2343%