NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-02-01M1GV5PG879N7CDJRGT747V86
Run Details
Model:
Z.AI GLM 5.3 (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
11m 20s
Cost:
$0.126938
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
99.92%
Matches word count
(details)
20.01%
Dialogue to Total Word Ratio
(details)
59.9652%
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 lay belly-down beneath a picnic table in Maple Grove Park, three houses down from the target, binoculars pressed to his face. Beside him sat a clipboard holding forty-seven pages of reconnaissance, including a hand-drawn map of the Hendersons' backyard with the trampoline marked OBSTACLE and the kiddie pool labeled WATER HAZARD. "Target is a six-year-old's birthday party," he murmured into his walkie-talkie. "Twelve children on site." Molly sat on the bench above him, eating a hot dog with alarming calm. "Why are we whispering? We're in a park." "Operational security." He clicked through photos on his phone: a unicorn piñata, a cupcake table, a pyramid of presents beneath a banner reading HAPPY BIRTHDAY, SOPHIE. According to his notes, the pile contained twenty-two gifts worth, conservatively, eight hundred dollars. "You counted presents at a birthday party. From under a picnic table." "From a strategic position." Bill unrolled a blueprint from his backpack—actual blueprint paper, cyan and creased, as though he'd robbed a 1950s architecture firm to get it. "Phase one, we neutralize the clown." "You can't neutralize a clown. He's doing balloon animals." "Nobody watches a clown. That's how clowns work." He tapped the blueprint. "Phase two, I create a diversion while you rappel from that oak." "There's no oak tree." Molly took a long bite of her hot dog. In nine years together they had lifted a Monet from a museum in Budapest, emptied a vault in Macau, and once stolen a billionaire's yacht while the man was asleep on it. She felt this was relevant context. "There are museums we're banned from. And you want to rob a six-year-old." "Their father installs security vaults. This is a message." "It's a birthday party." "It's victimless. Nobody remembers being six." "That's because of people like you." A tiny girl in a tutu wandered over from the party, dragging a foam sword, and studied Bill's camouflage pants sticking out from under the table. "Are you the magician?" Bill went rigid. Molly leaned down, smiling. "We're between shows, sweetheart." The girl nodded with terrifying seriousness and left to decapitate a dandelion. Bill exhaled and rallied, launching back into a timeline that had somehow acquired phases three through nine overnight. "Bill." Molly set down her hot dog, which, he now noticed, was topped with the Hendersons' signature jalapeño relish he'd documented on page twelve. She lifted the table's checkered cloth, revealing two garbage bags stuffed with wrapping paper and one very deflated unicorn piñata. "Signed us in as caterers. Walked out with everything while you drew an oak tree that doesn't exist." Bill stared at the bags. Then at her. Then at his blueprint, which suddenly looked less like a heist schematic and more like a craft project. "The clown?" he managed. "Flashed him the apron. Waved me right through." She stood, gathered the bags, and shouldered them like Santa Claus with a criminal record. "You can keep the walkie-talkie." From somewhere behind them, faintly, twelve children began singing happy birthday to an empty table.
created_at
"2026-09-02T10:38:26.32Z"
9/2/2026, 10:38:26 AM
model
"z-ai/glm-5.3-20260816"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1015
moderation_latency
(null)
generation_time
679643
tokens_prompt
75
tokens_completion
24912
native_tokens_prompt
73
native_tokens_completion
28843
native_tokens_completion_images
(null)
native_tokens_reasoning
28141
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
0.12693844
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788345506-jKws2mhsuSPtUXNSe65f"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788345506-FJ0fv3ngeNYtOEv4Kej0"
upstream_id
"20260902183826c54c7e7be63442f3"
provider_responses
0
endpoint_id
"db975786-e7f6-4838-9e56-3e0b77f1bdc2"
id
"20260902183826c54c7e7be63442f3"
is_byok
false
latency
1015
model_permaslug
"z-ai/glm-5.3-20260816"
provider_name
"Z.AI"
status
200
total_cost
0.12693844
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
99.92%
Matches word count
n/a
neededClean
false
words
503
20.01%
Dialogue to Total Word Ratio
Ratio: 35.67%, Deviation: 5.67%
neededClean
false
wordsTotal
513
wordsDialogue
183
59.9652%