NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-15-01KKS2QESD6R5VDFWT0DNPVF8H
Run Details
Model:
Z.AI GLM 5 Turbo
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
4m 1s
Cost:
$0.087583
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
99.95%
Dialogue to Total Word Ratio
(details)
99.9755%
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 unrolled the crinkled architectural blueprint across the greasy diner table, using a half-eaten ketchup packet as a paperweight. The document detailed the suburban fortress they were targeting, complete with a sprawling backyard and the most dangerous obstacle of all, a bouncing castle. Bill leaned back in the vinyl booth, nervously chewing on a toothpick. The sheer financial payout waiting inside that house was simply too massive to ignore. These were not ordinary toys. They were limited edition items, reselling for triple retail value. "Are we entirely sure about this?" "It is a high-yield target," Molly replied, tapping a black gloved finger on the diagram. "Twenty kids, all turning seven. That means at least twenty gaming consoles, fifty action figures, and a drone." "I am not talking about the payload," Bill whispered, leaning in closer so the waitress would not hear. "I am talking about the guards. They are unpredictable, highly caffeinated, and completely immune to reason." He had personally seen what a wild horde of excited graders could do to a grown man during a game of musical chairs, and the memory haunted his nightmares. Molly sighed loudly, pulling a bright red rubber nose from her heavy tactical vest pocket and softly placing it right on the center of the table. She carefully adjusted the tight bun in her dark hair and checked her watch. The infiltration window was between the pizza and the cutting of the cake, a notoriously chaotic fifteen minutes where parental supervision dropped to absolute zero. "That is exactly why we have the disguises. You are going in as a magician." "I do not know any magic." "You just need to wear the cape and hand out the smoke bombs if anyone gets too close to the present table." Bill picked up the sequined purple cape, holding the garment away from his body as if it were dangerously contaminated. He glanced out the window toward the residential street where their escape vehicle, a heavily modified minivan with a fake dent on the side, waited patiently. Stealing money from a heavily fortified downtown bank was one thing, but navigating a minefield of scattered Legos and sticky juice pouches required an entirely different skillset. "What happens if the real clown shows up early?" "The clown was cancelled. Budget cuts. We are clear." "And the piñata?" "I will handle the piñata. You just focus on bagging the gifts and moving them to the van. Remember, no sudden movements near the sugar rush zone." Bill nodded his head slowly, pulling the ridiculous purple cape over his broad shoulders. He looked absolutely absurd, a hardened criminal dressed like a birthday party entertainer, but this was why they were the best in the business. They had robbed a museum, a diamond exchange, and a heavily guarded politician's mansion, yet here they were, preparing to wage war against a troop of sugar-fueled children. Nobody suspects a magician of grand larceny at a seven-year-old's birthday bash. "Alright, let us go make some magic."
created_at
"2026-03-15T15:44:38.197Z"
3/15/2026, 3:44:38 PM
model
"z-ai/glm-5-turbo-20260315"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2337
moderation_latency
(null)
generation_time
241393
tokens_prompt
77
tokens_completion
21166
native_tokens_prompt
66
native_tokens_completion
27350
native_tokens_completion_images
(null)
native_tokens_reasoning
20386
native_tokens_cached
0
num_media_prompt
(null)
num_input_audio_prompt
(null)
num_media_completion
0
num_search_results
(null)
origin
"https://ncbench.com/"
is_byok
false
finish_reason
"stop"
native_finish_reason
"stop"
usage
0.08758336
router
(null)
provider_responses
0
endpoint_id
"9ce3b132-3a17-4fea-8092-0e8e519ed32b"
id
"2026031523443879d383288f174755"
is_byok
false
latency
2337
model_permaslug
"z-ai/glm-5-turbo-20260315"
provider_name
"Z.AI"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.0; linux; x64))"
http_referer
(null)
api_type
"completions"
id
"gen-1773589478-8X2lLGUeVOZoTwZyIdsW"
upstream_id
"2026031523443879d383288f174755"
total_cost
0.08758336
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Z.AI"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
500
99.95%
Dialogue to Total Word Ratio
Ratio: 29.90%, Deviation: 0.10%
neededClean
false
wordsTotal
505
wordsDialogue
151
99.9755%