NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-02-01M1GX885YYPER3EGM71KXZM4G
Run Details
Model:
Z.AI GLM 5.3 (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
5m 39s
Cost:
$0.071710
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
60.28%
Matches word count
(details)
94.66%
Dialogue to Total Word Ratio
(details)
77.4677%
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 was a napkin. Specifically, a napkin from the chip shop on Caldwell Road, with a gravy stain roughly where the bouncy castle would be, but Molly spread it across the van's dashboard with the solemnity of a general reviewing invasion plans. Beside her, Bill studied the wobbly rectangle marked TARGET and felt his stomach perform a slow, regretful roll. He'd pulled bank jobs, museum jobs, one long afternoon involving a llama, and never once had his nerves done this. There was something about robbing children that felt like punching downward. "Target's the rear garden," Molly said, tapping the stain. "Eleven hundred hours. That's when the magician goes on." "It's a kids' party, Molly." "Which is why nobody sees it coming." "Kids are witnesses, is what kids are. Hundreds of tiny witnesses with sticky hands and no concept of a false alibi." Molly ignored this, the way she ignored most things that were true. From the back seat she produced two aprons: one reading KISS THE COOK, the other WORLD'S OKAYEST DAD. Bill was handed the second and told not to ask questions. He asked questions anyway — about the exit, the take, and what happened if a child hugged him, which he insisted was documented caterer behavior. "The exit," Molly said, "is a wheelbarrow. We load the presents, wheel them out the front gate while the magician makes somebody disappear, and blend." "Blend with what? We're two adults stealing presents from a six-year-old." "We're catering staff, Bill. Nobody looks at catering staff." Bill looked across the street at number fourteen, already festooned — balloons on the mailbox, streamers on the hedge, a banner reading HAPPY 6TH TIMMY in lettering so cheerful it felt like a threat. Somewhere behind that hedge, according to the school-gate intelligence network, the presents were stacking up: remote-control cars, stuffed animals, and — prize of prizes — a bicycle from Timmy's grandmother. Not just any grandmother. Rosemary Wexler. Hotel Wexlers. A bicycle with gears, hydraulics, and a price tag Molly had once described, misty-eyed, as "organ money." "Suppose I go along with this," Bill said. "Suppose the plan works and we get out clean. What do we do with eleven presents shaped like dinosaurs?" "There are forty presents. I counted." "You counted?" "Callum counted. Callum is my rental child and a professional. He got in with a clipboard." Bill absorbed this. Somewhere in his chest, his conscience cleared its throat and was ignored. Outside, a balloon nudged the windscreen, tap-tap-tap, like a small ghost asking them to reconsider their lives. He looked down at the apron in his lap. WORLD'S OKAYEST DAD. It sat there like a prophecy. "Fine," he said. "But if a child hugs me, I'm out. That's my line." "There are no take-backs on a Wexler bike." He pulled the apron over his head. The van smelled of chips and destiny. Molly folded the napkin with the tenderness of a curator archiving a masterpiece, started the engine, and rolled them toward a suburban street where eleven six-year-olds were, at that very moment, unknowingly guarding the biggest score of their careers.
created_at
"2026-09-02T11:14:47.11Z"
9/2/2026, 11:14:47 AM
model
"z-ai/glm-5.3-20260816"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3224
moderation_latency
(null)
generation_time
339240
tokens_prompt
75
tokens_completion
13846
native_tokens_prompt
73
native_tokens_completion
16291
native_tokens_completion_images
(null)
native_tokens_reasoning
15578
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"
service_tier
(null)
usage
0.07170964
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788347687-QBrwTt5QvwbeolK9L0Wu"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788347687-2TpFU8MOXxRVmE8RMizm"
upstream_id
"20260902191447bb1e35b7ef174266"
provider_responses
0
endpoint_id
"db975786-e7f6-4838-9e56-3e0b77f1bdc2"
id
"20260902191447bb1e35b7ef174266"
is_byok
false
latency
3224
model_permaslug
"z-ai/glm-5.3-20260816"
provider_name
"Z.AI"
status
200
total_cost
0.07170964
cache_discount
0.00007296
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
60.28%
Matches word count
n/a
neededClean
false
words
515
94.66%
Dialogue to Total Word Ratio
Ratio: 31.05%, Deviation: 1.05%
neededClean
false
wordsTotal
525
wordsDialogue
163
77.4677%