NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-02-01M1GW97ZPAB97MHRBSGQS384C
Run Details
Model:
Z.AI GLM 5.3 (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
5m 55s
Cost:
$0.138614
Provider:
Z.AI
Evaluation summary
Score
Evaluator
Details Link
3.62%
Matches word count
(details)
67.60%
Dialogue to Total Word Ratio
(details)
35.6110%
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
Molly unrolled the blueprints across the kitchen table, pinning the corners with bags of confiscated candy and a stapler stolen from a post office. Around her, the kitchen had been transformed: photos of the cul-de-sac on the fridge, red string connecting the bouncy castle to a prosecco-wielding mother, and one manila folder labeled, in ominous capitals, PIPER. "The target," she said, tapping the paper. "Piper Wexler's seventh birthday. Saturday, two p.m. Thirty guests. Bouncy castle in the rear garden." Bill squinted at the blueprints, which had been drawn in crayon. "This is a floor plan of a semi-detached house." "I drew it from memory. I was at the Easter egg hunt." "You were invited?" "I'm very charming." Bill leaned closer, careful not to disturb the load-bearing string. "What's the red circle in the lounge?" "Present table. Sixty-one gifts, give or take. Aunt Carol over-shops." Molly traced the entry route with a lacquered nail. "We come through the side gate during the magician. Nobody watches the doors during the magician." "And the parents?" "Distracted. There's a prosecco fountain." Bill frowned. He'd robbed banks with better security and worse prosecco. "Last job, the toddler saw you." "The toddler was blind." "She had eyes, Molly. She pointed at you and said 'lady.'" "Then she fell over. Toddlers fall over constantly. It proves nothing." Bill paced to the window, arms folded in the posture of a man who believed he was the sensible one. Outside, a neighbor's cat watched him with unwarranted suspicion. "And the bouncy castle?" he said. "That's our escape route." "I've accounted for it." Molly slid over a second sheet: an annotated diagram complete with wind speeds, weight limits, and a small warning drawn at its center. "We go around." "Revolutionary." "Don't sass me. I did recon." "You stood at the fence taking notes while a seven-year-old asked to show you her cartwheels." "And I said yes, because I'm not a monster, and it confirmed the layout." She jabbed the blueprints. "Focus. Phase three: exfiltration. We load presents into the van—" "A van? You're reversing a plain white van into a cul-de-sac full of mums?" "It's a party van." "It's a trap. Nobody brings a van to a party except people who take things away from parties." Molly paused, pen hovering. He had a point. Infuriating, but a point. "Bunting," she said at last. "What?" "We decorate it. Balloons, streamers. 'Party Van Entertainment Service.' Nobody questions a van with bunting." Bill stared at her. On the fridge, Piper's school photo beamed between the red string, gap-toothed and oblivious to the storm gathering around her loot. "That's the stupidest plan I've ever heard," he said. "Including the spoon tunnel." "The spoon tunnel was structurally sound." "We're still paying for the spoon tunnel." Molly squared the blueprints into a crisp stack, because details were what separated professionals from people who simply grabbed presents and sprinted — which, she conceded, remained the backup plan. "Saturday, two p.m.," she said. "Wear something cheerful." Bill sighed the sigh of a man who had once dreamed of diamonds. "I'm not dressing as a clown." "You're dressing as a clown a little bit."
created_at
"2026-09-02T10:57:51.102Z"
9/2/2026, 10:57:51 AM
model
"z-ai/glm-5.3-20260816"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
3601
moderation_latency
(null)
generation_time
355172
tokens_prompt
75
tokens_completion
26420
native_tokens_prompt
73
native_tokens_completion
31480
native_tokens_completion_images
(null)
native_tokens_reasoning
30728
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
0.1386142
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788346671-pRd13fOAdRHrqivCnq7X"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788346671-JO6lyEPNTZCyiILi8UEH"
upstream_id
"20260902185751dda09b583f0f4458"
provider_responses
0
endpoint_id
"db975786-e7f6-4838-9e56-3e0b77f1bdc2"
id
"20260902185751dda09b583f0f4458"
is_byok
false
latency
3601
model_permaslug
"z-ai/glm-5.3-20260816"
provider_name
"Z.AI"
status
200
total_cost
0.1386142
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
3.62%
Matches word count
n/a
neededClean
false
words
524
67.60%
Dialogue to Total Word Ratio
Ratio: 52.80%, Deviation: 2.80%
neededClean
false
wordsTotal
536
wordsDialogue
283
35.6110%