NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-07-09-01KX40FBBK45EPZ0PBCP3VQ1TM
Run Details
Model:
GPT-5.6 Sol (Reasoning)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
34.7s
Cost:
$0.082890
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
95.99%
Matches word count
(details)
92.43%
Dialogue to Total Word Ratio
(details)
94.2065%
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
At precisely three o’clock, Molly unfolded a crayon map of the Buttercup Community Hall across the bonnet of their getaway vehicle, a squeaking ice-cream van painted with a penguin wearing handcuffs. Inside, Bill wrestled a balloon arch that had somehow inflated around his waist and pinned both elbows firmly backward. “First problem: you’re dressed as a burglar.” “I’m dressed as a professional.” “You’re wearing a striped shirt, Bill.” “It breathes.” “And the mask?” “Sun protection.” “It’s February.” “Winter sun is notoriously judgmental.” “Take it off. We enter as entertainers, locate the gift table, and leave quietly before cake is cut.” Bill peeled away the mask, revealing a second, smaller mask drawn in eyeliner. Molly stared until he scrubbed it off with a napkin advertising frozen peas. Their disguises waited nearby: two glittering capes, matching top hats, and a suitcase labelled MARVELLOUS MOLLY AND ADEQUATE BILL in optimistic gold paint prominently. “Why am I only adequate?” “Because ‘Magnificent Bill’ wouldn’t fit.” “You could’ve used smaller letters.” “I did.” “What’s our act?” “Disappearing presents.” “That feels less like an act and more like evidence.” “We’ll perform one trick first.” “Which trick?” “The one where you stop asking questions.” “That never works.” They entered through the kitchen, where forty jelly cups trembled like nervous witnesses. Beyond the swinging doors, children thundered around a papier-mâché dragon while exhausted adults pursued them with damp wipes. The present table stood beneath a banner reading HAPPY SEVENTH BIRTHDAY, guarded by one suspicious grandmother and several cupcakes. “Grandmother at two o’clock.” “That’s a chair.” “Grandmother behind the chair.” “She’s asleep.” “She’s knitting.” “With her eyes closed?” “Experts don’t need eyes.” “Neither do potatoes.” “Focus, Bill. Start juggling.” “I can’t juggle.” “Then drop things rhythmically.” “I trained for this.” “You didn’t.” Bill marched into the hall and removed three oranges from his hat. He tossed the first upward, the second sideways, and the third directly into a bowl of custard. Every child cheered. Encouraged, he bowed, backed into the dragon, and vanished beneath its enormous purple bottom completely from public view. Molly glided toward the presents, flourishing her cape. The grandmother opened one eye. Molly produced a wand, tapped a parcel, and whispered nonsense with such authority that three parents applauded. Then she swept five packages into the false lining of her cape, which immediately sagged to her ankles with bells. The bells jingled. The grandmother rose. Bill, still trapped inside the dragon, attempted a diversion by galloping across the hall. Unfortunately, he galloped into the buffet, acquired a tablecloth like a bridal train, and dragged sausage rolls behind him in a neat, incriminating trail toward the kitchen exit door nearby. Molly seized the dragon’s tail and steered. Together they burst through the kitchen, scattered jelly cups, and reached the van. Only then did Molly inspect their haul: five beautifully wrapped boxes containing socks, a multiplication workbook, two wooden recorders, and a craft kit. It burst open, coating both criminals silver.
created_at
"2026-07-09T17:58:10.556Z"
7/9/2026, 5:58:10 PM
model
"openai/gpt-5.6-sol-20260709"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
29828
moderation_latency
(null)
generation_time
33928
tokens_prompt
77
tokens_completion
941
native_tokens_prompt
66
native_tokens_completion
2752
native_tokens_completion_images
0
native_tokens_reasoning
2070
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
0.08289
router
(null)
provider_responses
0
endpoint_id
"a54c5de0-89bf-4ad7-a212-cf977eed918a"
id
"resp_09a902357bb69ec2016a4fe132a11481939caf747aa77b3868"
is_byok
false
latency
420
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-1783619890-HBO95enyjUSKo0Bw9JPI"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1783619890-QtaPsa4B1tPhvKzUcfMV"
upstream_id
"resp_09a902357bb69ec2016a4fe132a11481939caf747aa77b3868"
total_cost
0.08289
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
95.99%
Matches word count
n/a
neededClean
false
words
492
92.43%
Dialogue to Total Word Ratio
Ratio: 28.74%, Deviation: 1.26%
neededClean
false
wordsTotal
494
wordsDialogue
142
94.2065%