NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1M16BEX3P538EZHA322HQRM
Run Details
Model:
Qwen 3.8 27B (Reasoning, XHigh)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
2m 44s
Cost:
$0.033077
Provider:
Ionstream
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
21.37%
Dialogue to Total Word Ratio
(details)
10.6847%
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 and Bill sat on a rooftop above a suburban backyard, where streamers snapped in the wind and plastic frogs lay among the grass. The party below glowed with neon cakes, bouncy floors, and a clown who practiced jokes at a paper table for one hour while watching the children. "We must steal every single present from the party tonight." "The children will suspect us if we wear silly capes." "We enter through the plastic bounce castle very slowly, Bill." "The cake has sprinkles that look like really tiny clues." "We map every balloon and every candy table very carefully." Bill adjusted his duck mask, which had one eye hole too low and a beak that twitched when he worried. Molly checked her glitter gloves, her balloon hair, and the map she had drawn on a napkin during lunch as birds watched from the fence and laughed without noise again. "A duck costume is our absolutely perfect disguise tonight, Molly." "The parents are watching cartoons on their glowing little phones." "We hide all the presents inside the giant pink pinata." "If the clown wakes, you distract him with magic bubbles." "Your glitter gloves will make every stolen present sparkle beautifully." The party parents argued about playlist volumes near the punch table, while a tablecloth fluttered like a surrender flag. Bill pointed at the pinata, and Molly nodded, pretending to admire a lollipop the size of a beach ball as a duck costume waited nearby in the shade of balloons bright. "We take the shiny toy robot first, of course, Bill." "The punch cup has a bright silver spoon inside it." "We roll all the presents under the long juice table." "A toddler might ask us why the cake is gone." "We tell him it became a very fast flying present." The clown began a magic act involving a rubber fish and a napkin full of impossible wishes. Bill slipped behind the punch table, while Molly arranged the laser hat and counted the candy bags like a banker counting coins under a sign that said no running at the corner today. "Your balloon hair will convince the party photographer completely, Molly." "We swap the party hat for a small laser hat." "The laser hat points very politely at every gift bag." "We load all the presents into the striped candy van." "The van is decorated with lots of tiny smiling stars." By the time the last gift disappeared, the candy van hummed beside a hedge, wearing a fake police sticker and a bumper note reading please enjoy your free smile. The thieves drove away slowly, leaving only glitter, a rubber fish, and one suspiciously empty pinata, and many balloons and confetti.
created_at
"2026-09-03T16:21:22.535Z"
9/3/2026, 4:21:22 PM
model
"qwen/qwen3.8-27b-20260814"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
242
moderation_latency
(null)
generation_time
164080
tokens_prompt
75
tokens_completion
11764
native_tokens_prompt
118
native_tokens_completion
12955
native_tokens_completion_images
(null)
native_tokens_reasoning
12406
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
0.03307655
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788452482-HVwIGdOd3pIRD04mVxaf"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788452482-vumuvsm7AdV90ZahO1QA"
upstream_id
"f2ca967258da4fe4bbcfc8ad8595a596"
provider_responses
0
endpoint_id
"360e373e-d445-4de9-ae30-e33d6198715b"
id
"f2ca967258da4fe4bbcfc8ad8595a596"
is_byok
false
latency
242
model_permaslug
"qwen/qwen3.8-27b-20260814"
provider_name
"Ionstream"
status
200
total_cost
0.03307655
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Ionstream"
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
450
21.37%
Dialogue to Total Word Ratio
Ratio: 44.44%, Deviation: 5.56%
neededClean
false
wordsTotal
450
wordsDialogue
200
10.6847%