NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-24-01KQ05X04Z446HBBG6Q6NSJTPD
Run Details
Model:
Z.AI GLM 5.1
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
9m 6s
Cost:
$0.071590
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
98.08%
Dialogue to Total Word Ratio
(details)
98.9999%
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 adjusted the binoculars, zooming in on the sprawling suburban backyard of 42 Maple Drive. The target was a towering, princess-themed bounce house, inflated to maximum capacity and swarming with sugared-up, hyperactive six-year-olds. Bill sat next to her in the rusted-out van, unrolling a blueprint that turned out to be the instructions for a Lego castle. "This is a fortress, Molly. Look at the perimeter. Snack tables at three and six o'clock. A hired clown near the sandbox. And the mother? She’s got eyes like a hawk." Molly slapped the paper out of his hands, her patience wearing thinner than the rusted van's floorboards. She pointed a black-gloved finger at the brightly colored obstacle course occupying the center of the yard, treating the inflatable monstrosity like a fortified bank vault. "Forget the clown. That bounce house is our way in. It’s a structural anomaly. They won't expect us to enter through the mesh ceiling." Bill stared at her, his bushy eyebrows practically vanishing under his wool ski mask. He pulled a stray piece of pink glitter off his black turtleneck, looking deeply offended by the sheer existence of the operation. The van smelled like stale coffee and desperation, a potent combination for a duo used to cracking high-end museum safes. "Breaching the bouncy castle? We’re thieves, not acrobats! What about the tactical approach? We go through the side gate, neutralize the piñata, and grab the loot." Molly sighed heavily, reaching into the dusty glove compartment and pulling out a ziplock bag filled with sugar-free juice boxes. She handed one to him, her expression entirely serious, as if she were handing over a military-issue canteen before deploying into hostile territory. "The side gate is booby-trapped. I lost loafers to a Slip 'N Slide. We use the bounce house to gain elevation, drop down during cake cutting, and grab presents. Foolproof." Bill looked down at the juice box, then back up at the yard. A high-pitched squeal erupted from the children, causing him to flinch violently back against the seat. He gripped his steering wheel defensively, his eyes darting across the lawn, scanning for any sign of a parental ambush. "They’re vicious. Timmy Perkins bit a guy last Tuesday over a ring pop. What if they catch us? I’m not doing time in the timeout corner." Molly checked her waterproof watch. The ice cream truck was due in exactly five minutes, which would provide the perfect auditory cover. She tapped the plastic straw against the dashboard, mentally timing the upcoming melody and preparing for the distraction. "Then we don't get caught. Drink your camouflage. We move on my signal." Bill punctured the foil cap with the straw, scanning the perimeter with renewed determination. The sugar water hit his system immediately, replacing his earlier apprehension with a burst of artificial energy. He adjusted his ski mask, ignoring the brutal summer heat, and pulled the van door handle. He was ready to face the horde. The heist was on, and no toddler would stand in their way.
created_at
"2026-04-24T16:41:52.051Z"
4/24/2026, 4:41:52 PM
model
"z-ai/glm-5.1-20260406"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
421
moderation_latency
(null)
generation_time
545673
tokens_prompt
77
tokens_completion
18329
native_tokens_prompt
66
native_tokens_completion
20450
native_tokens_completion_images
(null)
native_tokens_reasoning
17541
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
0.07159022
router
(null)
provider_responses
0
endpoint_id
"e888c44c-6a07-4a9e-bc56-cb26511865ec"
id
"chatcmpl-RalVul4Awqhb1bp1UuLyc8vP"
is_byok
false
latency
94
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-1777048912-kT8bTRCyhJ7xdsejsyqm"
session_id
(null)
api_type
"completions"
id
"gen-1777048912-7V9D3E7dh5iP1FCRElzH"
upstream_id
"chatcmpl-RalVul4Awqhb1bp1UuLyc8vP"
total_cost
0.07159022
cache_discount
0.000054079
upstream_inference_cost
0
provider_name
"DeepInfra"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
99.92%
Matches word count
n/a
neededClean
false
words
503
98.08%
Dialogue to Total Word Ratio
Ratio: 29.38%, Deviation: 0.62%
neededClean
false
wordsTotal
514
wordsDialogue
151
98.9999%