NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ0Q226BXSR3Q7658VQKFJ1E
Run Details
Model:
GPT-5.5
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
10.2s
Cost:
$0.010050
Provider:
OpenAI
Evaluation summary
Score
Evaluator
Details Link
90.48%
Matches word count
(details)
36.79%
Dialogue to Total Word Ratio
(details)
63.6358%
Overall
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
“Bill, that’s not a vault.” “It’s a bouncy castle, Molly. Same principle: air pressure, guards distracted, treasure inside.” “The treasure is crayons, socks, and possibly a tambourine.” “Wrapped treasure. We don’t judge packaging.” “We are not stealing from children.” “We’re borrowing mystery boxes from a highly sugared crowd.” “Bill.” “What?” “The birthday boy is six.” “Then he won’t appreciate the artisanal ribbon work.” “He’ll appreciate his presents.” “Fine. We don’t steal all the presents.” “We steal none of the presents.” “What about one present?” “No.” “What about the clown’s wallet?” “No.” “The clown looks suspicious.” “The clown is tying balloon giraffes.” “That’s what he wants you to think.” “Bill, the plan is off.” “Already? I drew a map.” “That’s a napkin with ‘cake’ circled eight times.” “Cake is strategic.” “Molly, look. We sneak in as entertainers, assess the gift table, then dramatically reform.” “Reform?” “Yes. We announce we’re retired thieves and teach the children about ethics.” “With what qualifications?” “Experience.” “That’s the worst idea you’ve had.” “But wholesome.” “Wholesome doesn’t require lockpicks.” “These are party-favor tweezers.” “Put them away.” “Can we at least take cake?” “If they offer.” “And a balloon sword?” “If the clown permits it.” “Then this heist has become a luncheon.” “Exactly.” “I hate growth.” “Try frosting.”
created_at
"2026-04-24T21:41:43.767Z"
4/24/2026, 9:41:43 PM
model
"openai/gpt-5.5-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
725
moderation_latency
(null)
generation_time
10093
tokens_prompt
77
tokens_completion
492
native_tokens_prompt
66
native_tokens_completion
324
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"
usage
0.01005
router
(null)
provider_responses
0
endpoint_id
"58e5b336-423e-430b-a2ab-8bc353f0c51b"
id
"resp_09316872ed82709b0169ebe397eb7081948876cb7adadb8419"
is_byok
false
latency
319
model_permaslug
"openai/gpt-5.5-20260423"
provider_name
"OpenAI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777066903-iyGKi44NVEHje9A2TF5u"
session_id
(null)
api_type
"completions"
id
"gen-1777066903-S9kzzoeHjDLoCfHsBhTC"
upstream_id
"resp_09316872ed82709b0169ebe397eb7081948876cb7adadb8419"
total_cost
0.01005
cache_discount
(null)
upstream_inference_cost
0
provider_name
"OpenAI"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
90.48%
Matches word count
n/a
neededClean
false
words
210
36.79%
Dialogue to Total Word Ratio
Ratio: 100.00%, Deviation: 10.00%
neededClean
false
wordsTotal
211
wordsDialogue
211
63.6358%