NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-02-01M1GSN0DC3A2PH38K98MXG68T
Run Details
Model:
Z.AI GLM 5.3 Flash (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
12m 13s
Cost:
$0.009946
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
86.38%
Matches word count
(details)
11.65%
Dialogue to Total Word Ratio
(details)
49.0158%
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
Bill unveiled the blueprint at midnight, spreading it across the card table and weighting the corners with four cold coffee mugs. It was drawn in crayon. It was, Molly had to admit, thorough. "Tyler Henderson," he said. "Turning eight. Saturday, two p.m." She studied the map of a suburban backyard: a folding table marked PRESENTS, a swimming pool shaped like a whale, and a bouncy castle circled three times in red. Bill had drawn arrows everywhere. Molly, who had once escaped a diamond exchange through a laundry chute, considered crayon a new low. "Obstacles," Bill said, tapping the page. "A bouncy castle. A clown named Bubbles. And a dog called Bartholomew — twelve, mostly deaf, and technically alive." "The presents?" "Twenty-seven, wrapped, unguarded. We go in during the piñata, when every kid goes blind with candy fever. Flawless." Molly lit a cigarette off the toaster, because the safehouse was technically a garage and the stove was technically a hazard. "Bill. We're planning a commando raid on a child's birthday." "We're planning a retrieval of unattended goods. It's different. It's legal-adjacent." The job had come through a guy Bill knew in Cleveland, who wanted the presents gone by Saturday for reasons of his own. Professional courtesy meant nobody asked twice. Outside, their getaway van — a rusted Econoline wearing the fake logo of a pet-grooming business — ticked as it cooled. Somewhere across town, forty children slept, unaware their candy-assorted futures hung in the balance. "The piñata is a load-bearing element of this plan," Molly said, "isn't it?" "The piñata is the whole plan." She exhaled smoke and studied the crayon square marked PRESENTS. Twenty-seven units. She had robbed banks with fewer moving parts. Banks never had clowns. "One condition," she said. "We don't touch the cake." "The cake?" The kid had talked about that cake for a month, apparently. Dinosaur theme. "You don't steal a boy's cake, Bill. Presents, fine. Cake is where I draw the line." Bill nodded with the gravity of a man accepting terms at a summit. "Cake stays. The bike box comes with us." "That's the fence's payment?" "That's the retirement plan." Molly pulled out a chair and sat with the resignation of a woman who had hidden in her share of ball pits. Across the table, Bill inventoried his gear: rope, suction cups, a smoke bomb, and two paper hats, one of which read HAPPY BIRTHDAY in glitter. "The plan is hats," she said. "Hats, piñata, Bubbles, who answers to no one," Bill said, folding the blueprint into his jacket with the grin that had gotten them banned from three state fairs and one funeral. "Nobody's ever robbed a kid's party and lived to spend the money." "Not the bragging point you think it is." Saturday came, and with it a sky the color of wet cardboard. The two of them stood at the Hendersons' back fence in matching paper hats, watching through the slats as Bubbles the clown twisted a balloon dachshund and Bartholomew slept through his own legacy. Molly checked her watch. "Two p.m.," Bill whispered. "Places."
created_at
"2026-09-02T10:11:50.838Z"
9/2/2026, 10:11:50 AM
model
"z-ai/glm-5.3-flash-20260826"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3438
moderation_latency
(null)
generation_time
733306
tokens_prompt
75
tokens_completion
32437
native_tokens_prompt
73
native_tokens_completion
39763
native_tokens_completion_images
(null)
native_tokens_reasoning
39048
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
"stop"
service_tier
(null)
usage
0.009946225
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788343910-lmTsQcc6OUAa0UxZqnde"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788343910-w0ATkGTNnpjgCpid571Q"
upstream_id
"202609021811517512d1bf49674713"
provider_responses
0
endpoint_id
"8e9fe48b-2f91-41c3-a8a7-e4a93a8c4ff0"
id
"202609021811517512d1bf49674713"
is_byok
false
latency
3438
model_permaslug
"z-ai/glm-5.3-flash-20260826"
provider_name
"Z.AI"
status
200
total_cost
0.009946225
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Z.AI"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
86.38%
Matches word count
n/a
neededClean
false
words
511
11.65%
Dialogue to Total Word Ratio
Ratio: 36.56%, Deviation: 6.56%
neededClean
false
wordsTotal
517
wordsDialogue
189
49.0158%