NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ060257854JPS47QW09J93Z
Run Details
Model:
Z.AI GLM 5.1
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
7m 14s
Cost:
$0.055787
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
68.67%
Dialogue to Total Word Ratio
(details)
84.3362%
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
Molly spread the heavily creased, crayon-drawn map across the hood of the stolen station wagon. The bright afternoon sun glared off the purple wax markings that depicted the formidable defenses of four-year-old Timmy Henderson's birthday bash. She adjusted her earpiece. The stakes were high. Inside that suburban fortress lay a pile of presents that would make a fence weep with joy. If they failed, they would be dealing with the terrifying wrath of a horde of sugar-crazed kindergartners and their even more terrifying mothers. "The target is heavily fortified," Molly said, tapping a drawing of a bouncy castle. "Inflatable perimeter is tight. Only one point of entry, and it is guarded by a clown." Bill squinted at the map, adjusting his black turtleneck despite the sweltering heat. Sweat beaded on his forehead. He gripped the handle of the station wagon, knuckles white with anticipation. "What kind of clown? Are we talking sad hobo clown or angry fireman clown?" "Mime," Molly corrected grimly. "Worse," Bill muttered, checking the pressure gauge on his customized glitter bomb. The canister gleamed under the sunlight, ready to deploy a tactical cloud of magenta sparkles. It was a non-lethal measure, designed to blind and disorient their diminutive adversaries without causing permanent harm. Collateral damage was to be avoided at all costs. No one wanted a repeat of the great Easter egg hunt disaster of last spring. "We go in at fourteen hundred hours. Right during the cake cutting. That is when their guard is down," Molly explained, tracing a route past the snack table. "I will trigger the bubble machine. You slip past the piƱata and hit the loot pile." "What about the parents?" Bill asked. "Distracted by the magician. You have a thirty-second window to grab the gift bags and get out." Bill chewed his lower lip, staring intently at the gingerbread man. "Are we taking the presents wrapped or unwrapped? Unwrapping on site takes time. We could lose precious seconds on the Lego sets." "We take the loot as is. We can fence the wrapping paper later," Molly declared. Bill nodded slowly, pulling a black ski mask over his face. The wool muffled his breathing and trapped the heat, making him look like a deranged winter ninja in the middle of summer. He reached into the backseat, pulling out a camouflage net custom-tailored for a tricycle. "I do not like this," Bill mumbled through the wool. "Last time we hit a toddler jam, I got a plastic saxophone jammed in my shin." "Casualty of war," Molly said, pulling her own mask down. "Now check your silly string. Safety off." Bill flipped the cap on his neon green canister. He gave the canister a firm shake, listening to the satisfying slosh of the pressurized foam inside. With a final nod to his partner, he stepped away from the vehicle, creeping toward the backyard gate. The sounds of loud, chaotic laughter and a bouncing bassline drifted over the wooden fence. "Safety off. Ready to party crash."
created_at
"2026-04-24T16:43:32.401Z"
4/24/2026, 4:43:32 PM
model
"z-ai/glm-5.1-20260406"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
588
moderation_latency
(null)
generation_time
433261
tokens_prompt
77
tokens_completion
13303
native_tokens_prompt
66
native_tokens_completion
15927
native_tokens_completion_images
(null)
native_tokens_reasoning
12531
native_tokens_cached
32
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
0.05578676
router
(null)
provider_responses
0
endpoint_id
"e888c44c-6a07-4a9e-bc56-cb26511865ec"
id
"chatcmpl-R9gTSf1TF50bi6Kdk3n5kvmL"
is_byok
false
latency
97
model_permaslug
"z-ai/glm-5.1-20260406"
provider_name
"DeepInfra"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777049012-1lkDh8zdqg6JngYnFA24"
session_id
(null)
api_type
"completions"
id
"gen-1777049012-4SQQhXmelRjGV48LcrZT"
upstream_id
"chatcmpl-R9gTSf1TF50bi6Kdk3n5kvmL"
total_cost
0.05578676
cache_discount
0.000027039
upstream_inference_cost
0
provider_name
"DeepInfra"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
500
68.67%
Dialogue to Total Word Ratio
Ratio: 32.74%, Deviation: 2.74%
neededClean
false
wordsTotal
507
wordsDialogue
166
84.3362%