NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1M0CGCNSEW3HRZ18JJ9FT2G
Run Details
Model:
Qwen 3.8 27B (Reasoning, XHigh)
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
2m 11s
Cost:
$0.026158
Provider:
Phala
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
16.33%
Dialogue to Total Word Ratio
(details)
8.1655%
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 spread a crayon map across the table, marking the kitchen window, the punch bowl, and the suspiciously cheerful clown. Her sequin blazer glittered, and a plastic star sat in her hair. Bill, across from her, polished a party horn. The room smelled of frosting, balloons, and small, confident crimes. "Molly, if we steal every present, the children will notice. But if we replace them with glitter, candles, and one rubber chicken, they will call us very clever magicians today." Molly traced a glittery line toward the snack table. The presents were arranged there like a row of tiny hostages, wrapped in bright paper that demanded trust. She pointed to the juice station, then to the bouncy castle, and to the clown, who was juggling three spoons with alarming success. "First, you create a distraction near the punch bowl. I will carry the largest gift to the backyard. Remember: the clown is not a threat, he is a mobile witness." Bill nodded, then held up the party horn as if it were a tiny brass promise. He had trained for this. He had practiced smiling while leaking air, and once terrified a dentist with a squeal. He believed in precision, timing, and the loud dignity of party noise. "You will blow once when the clown turns left. I will swap three gifts before he turns right. The bouncy castle is our tunnel, our shield, and our ridiculous exit strategy." Bill tapped the map with his elbow, leaving a frosting smear shaped like a crime scene. The presents glowed under the party lights, each ribbon a tiny accusation. Molly checked her watch, which had a plastic dinosaur wrapped around it. Bill straightened his tassel, which had already surrendered to gravity. "When the children start clapping, we move. When the clown says, 'Would anyone like to see my magic trick?' we leave through the backyard before the word 'magic' becomes legally binding." Molly smiled, because Bill was almost always right about the law, even when the law involved children, confetti, and a clown with suspiciously sharp eyebrows. Bill raised the party horn to his lips. The room seemed to hold its breath, as if the punch bowl had suddenly remembered it was supposed to be polite. "Now," Bill said, "let us go be charmingly criminal, before someone brings out a piƱata and we are forced to discuss our feelings."
created_at
"2026-09-03T16:07:15.611Z"
9/3/2026, 4:07:15 PM
model
"qwen/qwen3.8-27b-20260814"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
359
moderation_latency
(null)
generation_time
131176
tokens_prompt
75
tokens_completion
6791
native_tokens_prompt
186
native_tokens_completion
8702
native_tokens_completion_images
(null)
native_tokens_reasoning
8193
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
0.026158
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788451635-wb9I4f6hKkst5Qq0luqX"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788451635-Y7aErvdQxkdKhDg7tMmD"
upstream_id
"req_d6d24e18ed14d363fd1a2bea32f05778"
provider_responses
0
endpoint_id
"cfc4d0ca-df7c-4b5f-9c74-328c41a7cbc3"
id
"req_d6d24e18ed14d363fd1a2bea32f05778"
is_byok
false
latency
359
model_permaslug
"qwen/qwen3.8-27b-20260814"
provider_name
"Phala"
status
200
total_cost
0.026158
cache_discount
0.0000224
upstream_inference_cost
0
provider_name
"Phala"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
397
16.33%
Dialogue to Total Word Ratio
Ratio: 36.02%, Deviation: 6.02%
neededClean
false
wordsTotal
397
wordsDialogue
143
8.1655%