NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-23-01M37HFFEYAQW1HTG4MAH5WCW8
Run Details
Model:
GPT-6 Sol
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
17.3s
Cost:
$0.006772
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
51.93%
Matches word count
(details)
87.52%
Dialogue to Total Word Ratio
(details)
69.7239%
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 drawing across the café table. A purple rectangle occupied most of it. Around the edges, someone had written CAKE in increasingly frantic letters. “Tell me that’s a floor plan,” Bill said. “It’s a floor plan.” “Why does it have a sun wearing sunglasses?” “The birthday girl drew it. Her mother posted it online.” “Did she label the exits?” “She labeled the cake six times. The front door’s here.” Bill leaned closer. “That’s a carrot.” “It’s a door shaped like a carrot. They’re doing a rabbit theme.” “Then what’s the purple rectangle?” “Presents.” “You’re certain?” “Her mother posted seventeen photographs of the setup. The purple rectangle is a tablecloth.” Bill sat back. “I thought we were stealing from a children’s party.” “We are. Six-year-olds have rich grandparents.” “And parents with cameras.” “And a magician coming at two. We go in while everyone watches him pull something out of something else.” “What do we go in as?” Molly slid a second sheet across the table. Bill read it, then looked at her. “Balloon arch technicians?” “I rang the mother. She said the arch was already installed, so I told her we’d been sent to inspect it.” “By whom?” “The balloon company.” “Is there a balloon company?” “There is now. I printed invoices.” Bill studied the sheet. “Your company is called Arch Enemies.” “It made me laugh.” “It will make the police laugh.” “Then we’ll have brought some joy into a difficult day.” A waiter stopped beside them. Molly ordered tea. Bill ordered a bun, then waited until the waiter had gone. “How do we get the presents out?” he asked. “In those long balloon bags. Two each, straight through the side gate.” “All of them?” Molly glanced at her phone. “Forty-three packages so far. We’ll need several trips.” Bill stared at her. “Forty-three trips past a magician, twelve children, their parents and a rabbit-shaped front door?” “Not forty-three trips. Some are small.” “I’m reassured. What if a child asks where we’re taking them?” “We say we’re checking the gifts for balloons.” “That doesn’t mean anything.” “It doesn’t have to. We’ll be wearing lanyards.” Bill took another look at the floor plan. “Whose house is it?” “The Harcourts’. You know the old place behind the library?” “Of course I do. Big garden. Blue garage.” Molly frowned. “The garage is green.” “It was blue last week. I delivered a chest of drawers there.” She opened the photographs. In one, a blue garage stood behind a towering balloon arch. In another, a woman held a baby beneath a banner reading HAPPY FIRST BIRTHDAY, PIP. Molly enlarged the address on the invitation. “Oh.” Bill waited. “Next door,” she said. “The six-year-old lives next door.” “So we have no floor plan.” “We have a drawing.” “And an appointment to inspect the wrong balloon arch.” Molly folded the drawing. “We could steal the magician.” “What?” “Nothing. Tea first.”
created_at
"2026-09-23T16:27:14.789Z"
9/23/2026, 4:27:14 PM
model
"openai/gpt-6-sol-20260922"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3168
moderation_latency
(null)
generation_time
17271
tokens_prompt
75
tokens_completion
851
native_tokens_prompt
66
native_tokens_completion
664
native_tokens_completion_images
0
native_tokens_reasoning
0
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
0.006772
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.20.0; linux; x64))"
http_referer
(null)
request_id
"req-1790180834-9ZYIzpIvRGFrWKehXrxh"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1790180834-nO7QZaeQLGNVbdMGmLPq"
upstream_id
"resp_057892468d41520d016ab3fde2eb9c87d19a634247eeb3a0ba"
provider_responses
0
endpoint_id
"1cfc7d9d-4404-4b8e-9ee4-58ae45c9dcd4"
id
"resp_057892468d41520d016ab3fde2eb9c87d19a634247eeb3a0ba"
is_byok
false
latency
1155
model_permaslug
"openai/gpt-6-sol-20260922"
provider_name
"OpenAI"
status
200
total_cost
0.006772
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
51.93%
Matches word count
n/a
neededClean
false
words
484
87.52%
Dialogue to Total Word Ratio
Ratio: 71.63%, Deviation: 1.63%
neededClean
false
wordsTotal
490
wordsDialogue
351
69.7239%