NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-03-25-01KMJ8QQRH2XF4TJ0V6FS03A26
Run Details
Model:
Qwen3 235B A22B Instruct 2507
Test:
Dialogue tags
Scenario:
Write 200 words with 90% dialogue
Duration:
8.3s
Cost:
$0.000322
Provider:
AtlasCloud
Evaluation summary
Score
Evaluator
Details Link
35.00%
Matches word count
(details)
99.99%
Dialogue to Total Word Ratio
(details)
67.4983%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 90% 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: “You’re telling me we’re breaking into a birthday party? For kids?” Bill: “Exactly. Chaos, zero security, and a piñata full of cash? It’s a gold mine in clown form.” Molly: “So we dress as clowns?” Bill: “Nah, too suspicious. We go as disappointed uncles. Bring cheap beer. Lean against the minivan.” Molly: “I don’t know how to act like a disappointed uncle.” Bill: “Just breathe through your mouth and sigh a lot. Classic Uncle Dave energy.” Molly: “When’s the present opening?” Bill: “Ten minutes after cake, when everyone’s sugar-crashing and sticky. We swoop in during the ‘musical statues’ disaster.” Molly: “And if the parents ask what we’re doing?” Bill: “Say we’re with Child Protective Services. Bring clipboards. Kids love authority figures with clipboards.” Molly: “And the loot?” Bill: “Dump the gifts in the bounce house, climb in after. Bounce our way out.” Molly: “You’re insane.” Bill: “I’m inspired. Also, I already ordered the ‘Uncle Bubba’ name tags.” Molly: “Do we at least leave a balloon animal?” Bill: “Only if it’s filled with smoke pellets.” Molly: “Fine. But if I get glitter in my boots, I’m blaming you and that piñata.” Bill: “Worth it. Nothing says professional theft like stealing from people who still believe in Santa.” Molly: “That’s actually kind of sad.” Bill: “No, Molly. It’s *efficient*.”
created_at
"2026-03-25T10:30:39.639Z"
3/25/2026, 10:30:39 AM
model
"qwen/qwen3-235b-a22b-07-25"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
745
moderation_latency
(null)
generation_time
8290
tokens_prompt
77
tokens_completion
412
native_tokens_prompt
72
native_tokens_completion
350
native_tokens_completion_images
(null)
native_tokens_reasoning
0
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
0.0003224
router
(null)
provider_responses
0
endpoint_id
"aaae5602-0949-4bc9-bd77-588b4b285341"
id
"f776a08d55914a5ab87d32e705864c02"
is_byok
false
latency
745
model_permaslug
"qwen/qwen3-235b-a22b-07-25"
provider_name
"AtlasCloud"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.0; linux; x64))"
http_referer
(null)
api_type
"completions"
id
"gen-1774434639-K3FOL5KyK7Gi9NtHZB8O"
upstream_id
"f776a08d55914a5ab87d32e705864c02"
total_cost
0.0003224
cache_discount
(null)
upstream_inference_cost
0
provider_name
"AtlasCloud"
Evaluation details
Result
Evaluator
Details
Meta Data
35.00%
Matches word count
n/a
neededClean
false
words
218
99.99%
Dialogue to Total Word Ratio
Ratio: 90.87%, Deviation: 0.87%
neededClean
false
wordsTotal
219
wordsDialogue
199
67.4983%