NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHK9WTJXQNBM6EZG2DHC20E7
Run Details
Model:
GPT-5
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
1m 40s
Cost:
$0.068043
Provider:
OpenAI
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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
They huddled over a glittery party flyer; confetti drifted lazily while, somewhere nearby, a piñata squeaked with ominous cheer tonight. Molly: "All right, Bill, objective: liberate the loot." Bill: "I brought empathy, a kazoo, and glitter." Molly: "Kazoo?" Bill: "Diversion. Children hear jazz; adults hear defeat." Molly: "Codenames. You're Sir Wrappington." Bill: "Then you're Lady Bows-a-Lot." Molly: "We enter as volunteer gift-organizers, clipboards blazing." Bill: "Competence is my sneakiest costume." Molly: "We'll stack presents on the Wagon of Plausibility." Bill: "Does it squeak adorably?" Molly: "Like a hamster doing opera." Bill: "Contingency: if challenged, we audit sparkle levels." Molly: "And shuffle papers knowingly." Bill: "If chased, I deploy kazoo fugue." Molly: "Shout, 'Conga line compliance check!'" Bill: "Spell 'conga'?" Molly: "Incorrectly. Confidence outruns spelling." Bill: "Ethics inventory: are we monsters?" Molly: "Borrowers. We return the joy as confetti." Bill: "Post-heist snack protocol?" Molly: "Cupcakes, seized under the Friendship Clause." Bill: "Backup disguise?" Molly: "Balloon animal inspectors—strict about snakes." Bill: "I only make snakes." Molly: "Excellent. Minimalist poodles." Bill: "Extraction signal?" Molly: "When the cake knife sings." Bill: "Knives sing?" Molly: "In butter, every blade is Pavarotti." Bill: "Then onward, Lady Bows-a-Lot." Molly: "With charm, Sir Wrappington—no tears, no traces. Now."
created_at
"2026-02-16T13:24:44.106186+00:00"
2/16/2026, 1:24:44 PM
model
"openai/gpt-5-2025-08-07"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
12978
moderation_latency
(null)
generation_time
99595
tokens_prompt
77
tokens_completion
1341
native_tokens_prompt
66
native_tokens_completion
6796
native_tokens_completion_images
0
native_tokens_reasoning
6336
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"completed"
usage
0.0680425
router
(null)
provider_responses
0
id
"resp_08cf379e2866827e0169931a37f9b8819699fd6e5d1e1a0e26"
status
200
is_byok
false
latency
159
endpoint_id
"7c2f859a-7890-4e8e-b1de-1cd1c0a800b4"
provider_name
"OpenAI"
model_permaslug
"openai/gpt-5-2025-08-07"
api_type
"completions"
id
"gen-1771248183-240WtAs00gYO0yBS5eF0"
upstream_id
"resp_08cf379e2866827e0169931a37f9b8819699fd6e5d1e1a0e26"
total_cost
0.0680425
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
Evaluation details
Result
Evaluator
Details
Meta Data
95.9868%
Matches word count
n/a
neededClean
false
words
208
1.1749%
Dialogue to Total Word Ratio
Ratio: 75.48%, Deviation: 14.52%
neededClean
false
wordsTotal
208
wordsDialogue
157
48.5808%