NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-02-01M1GV2WPRJ64E6625BXZBHG42
Run Details
Model:
Z.AI GLM 5.3 (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
6m 31s
Cost:
$0.104674
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
93.65%
Matches word count
(details)
65.54%
Dialogue to Total Word Ratio
(details)
79.5943%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 50% 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 kitchen table wasn't a bank vault or a casino floor—it was a bounce house layout, drawn in crayon, with stick-figure guards at the snack station. Bill adjusted his reading glasses and tapped a wobbly rectangle labeled CASTLE OF FUN. "This is where we hit 'em. Twelve-thirty, entertainment arrives, parents cluster at the cake table. Gift pile sits undefended." Molly squinted at the drawing. There were heart doodles in the margins and a sun wearing sunglasses. "Did you draw this yourself?" "I interviewed the birthday girl." "You interviewed—Bill. She's six." "And she talked. Kids have no operational security whatsoever." He slid a photograph across the table: a mountain of wrapped boxes beneath a banner reading HAPPY BIRTHDAY MADISON. In the background, a man in cargo shorts wept over a grill. "Loot's worth eleven hundred, minimum. American Girl doll. A Nintendo. Some kind of robotic hamster—apparently the crown jewel." Molly rubbed her temples. Fifteen years of professional theft. Museum walls scaled. Diamonds smuggled inside a trombone. This was somehow worse. "We've done museums. We did that yacht. Now we're hitting a pizza party." "Nobody expects a heist at a bounce house." "Because it's insane!" Bill ignored her and unrolled a second document—a party itinerary, decorated with glitter glue. He pinned it flat with a saltshaker, tenderly, like a general handling battle maps. "One o'clock, magician. Our window. Every eye in the yard goes to Doug the Amazing and his dove-related nonsense." "Mom posted it online. Complete with gate code." Molly stared at him. "The gate code. On social media." "Four-seven-one-one. She also listed her work schedule and the fact that their dog is friendly. Honestly, I almost feel guilty." "You should feel guilty! These are children!" "Children whose parents registered for a ninja turtle pool float." He pointed at the corner of the blueprint. "You'll go in as the balloon artist." "We're not actually making balloons, are we?" Bill paused a beat too long. "Bill." "I watched four tutorials." "That's not a skill! You can't tutorial your way into balloon art!" Molly stopped, eyes narrowing at the duffel bag beside his chair. A rubber glove protruded from the zipper, next to something brightly, ominously colored. "Wait. You're putting me in a rainbow wig?" "There's also a novelty bow tie." "Two years in a Turkish prison, and this is what I get." "One job. In and out in nine minutes. You twist the dog, I load the wagon, gone before cake." Molly exhaled and looked down at the blueprint—specifically, the little stick figures Bill had drawn weeping over an empty present table. "You drew the kids crying." "That's the after-action projection. I thought it was thorough." "It's horrifying. We're changing the plan—" "There's a sequel," Bill said. "Twin party. Two towns over. July." Molly stood and snatched the wig from the duffel bag. She jammed it onto her head with the resignation of a woman who'd once rappelled off the Eiffel Tower in heels. "Fine," she muttered. "But I'm keeping the hamster."
created_at
"2026-09-02T10:36:54.367Z"
9/2/2026, 10:36:54 AM
model
"z-ai/glm-5.3-20260816"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1412
moderation_latency
(null)
generation_time
391438
tokens_prompt
75
tokens_completion
19064
native_tokens_prompt
73
native_tokens_completion
23783
native_tokens_completion_images
(null)
native_tokens_reasoning
23083
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
0.10467444
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788345414-Wa7CoqUPw2s4UZYdmDvf"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788345414-YdyU6En6qul8MXUa4UwU"
upstream_id
"202609021836544ee4bb28c1f5463b"
provider_responses
0
endpoint_id
"db975786-e7f6-4838-9e56-3e0b77f1bdc2"
id
"202609021836544ee4bb28c1f5463b"
is_byok
false
latency
1412
model_permaslug
"z-ai/glm-5.3-20260816"
provider_name
"Z.AI"
status
200
total_cost
0.10467444
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
93.65%
Matches word count
n/a
neededClean
false
words
509
65.54%
Dialogue to Total Word Ratio
Ratio: 52.91%, Deviation: 2.91%
neededClean
false
wordsTotal
516
wordsDialogue
273
79.5943%