NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-27-01KQ74MX6C929NPFKRKZGC3V13
Run Details
Model:
Qwen 3.6 27B
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
2m 15s
Cost:
$0.030647
Provider:
Morph
Evaluation summary
Score
Evaluator
Details Link
98.71%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
49.3562%
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
The blueprint spread across the chipped Formica table resembled a Rorschach test drawn by a maniac. Crayon marks bled into coffee rings and the sticky residue of last night failed attempt to steal a bakery worth of muffins. Molly traced the perimeter of the suburban cul de sac with a grease pencil, her jaw tight as she considered the tactical nightmare ahead. Bill sat on a mismatched lawn chair that groaned under his weight, adjusting a pair of night vision goggles that had belonged to a discarded department store return. He looked like a raccoon trying to pass as a tactical operator. The target was a seven year old birthday celebration for a child named Timothy who had never once asked for socks. Bill cleared his throat and straightened his posture on the wobbly chair. The clock ticked louder in the quiet room. They both knew tonight would change everything. We could definitely enter through the sliding glass door, Bill said, tapping the plastic wrap. The homeowner installed exactly two motion sensors and a very aggressive golden retriever. I vote we disguise ourselves as inflatable parade floats. You are suggesting we wear neon ponchos and ride tiny rocking horses into a suburban fortress because you are afraid of a single dog. A well-trained canine absolutely respects authority. Besides, the main event is the gift pyramid. We need extraction routes that account for sticky fingers and sudden emotional meltdowns. We are not getting paid to wipe tears, Molly replied. We are paid three hundred dollars in unmarked twenty-dollar bills to remove every plastic dinosaur and board game without triggering parental alarm. We hit the snack table on zero-four hundred hours, Bill murmured. I cover the left flank while you bypass the balloon arch and make straight for the gift table. She snatched up a stress ball shaped like a watermelon, squeezing it until the rubber squeaked in protest. Three pounds of disappointment and novelty plastic would not buy them a decent getaway vehicle. Yet here they were, treating a suburban afternoon like a midlife crisis. Bill produced a small clipboard and began scribbling notes about optimal sneaking trajectories near the fondue fountain. He treated the entire operation like a military campaign against a sovereign nation. Molly stared at him as if he had just suggested they tackle a live artillery battery with wooden spoons. She straightened her spine and smoothed the blueprint flat with both hands. The house sat too close to the neighbor property line, which meant any noise above a whisper would bring down the entire suburban watch committee. Still, she nodded slowly, a grim smile touching her lips as she pictured the mountain of wrapping paper. Then she grabbed her coat and reached for the car keys, ready to march toward whatever ridiculous fate awaited them next. The radio crackled with static and poorly recorded jazz music. Outside, a sprinkler system activated with military precision, drenching the lawn in perfectly timed bursts.
created_at
"2026-04-27T09:34:39.314Z"
4/27/2026, 9:34:39 AM
model
"qwen/qwen3.6-27b-20260422"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
701
moderation_latency
(null)
generation_time
134673
tokens_prompt
77
tokens_completion
10682
native_tokens_prompt
76
native_tokens_completion
12752
native_tokens_completion_images
(null)
native_tokens_reasoning
9925
native_tokens_cached
64
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
"stop"
usage
0.0306466
router
(null)
provider_responses
0
endpoint_id
"826cab4e-dc26-4472-83be-e4ddd5ebd2ab"
id
"695a6b68fb7c44bf8defd267194ad89f"
is_byok
false
latency
651
model_permaslug
"qwen/qwen3.6-27b-20260422"
provider_name
"Morph"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777282479-QbTT0Hdba8XCyphS4cTJ"
session_id
(null)
api_type
"completions"
id
"gen-1777282479-NXijaxTDtGAjdq238V1p"
upstream_id
"695a6b68fb7c44bf8defd267194ad89f"
total_cost
0.0306466
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Morph"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
98.71%
Matches word count
n/a
neededClean
false
words
494
0.00%
Dialogue to Total Word Ratio
Ratio: 0.00%, Deviation: 30.00%
neededClean
false
wordsTotal
497
wordsDialogue
0
49.3562%