NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-20-01M5C66T6YR7B7STT2CY5YCEVQ
Run Details
Model:
DeepSeek V4.1 Flash 1017 (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
56.7s
Cost:
$0.005044
Provider:
InferenceNet
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
96.99%
Dialogue to Total Word Ratio
(details)
98.4972%
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 unfolded the crayon-covered invitation across the van’s dashboard, smoothing it flat with the reverence usually reserved for bank blueprints. “Right, Bill. Saturday, two o’clock, Pirate Pete’s Play Palace. Thirty toddlers, two exhausted parents, and a mountain of presents in the party room.” Bill adjusted his balaclava in the rearview mirror and stared at her as if she had suggested robbing a police station during a parade. “Remind me why we’re robbing a six-year-old.” Molly tapped the invitation and grinned. “Because the museum job had lasers and this has cake. Also, the birthday boy’s father allegedly bought a solid-gold rocking horse.” “Allegedly is how you ended up in that clown suit last time.” “That clown suit saved us. Security is one teenager in a shark costume and a bouncy castle with a weight limit. We go in as balloon artists.” Bill wrote on a napkin with the grave concentration of a bomb defuser. “Two middle-aged men as balloon artists?” “No, two middle-aged men as balloon artists who happen to carry large empty sacks. You make a poodle; I make a distraction.” “I can’t make a poodle. I can make a snake that looks like a sausage.” “Perfect. Nobody questions sausage art. Now, the plan: we enter through the service door behind the ball pit, blend in during musical chairs, and when they sing happy birthday, we bag every gift with one swift sweep.” Molly leaned over and drew a crude map of the play palace, circling the ball pit, the cake table, and a door marked STAFF ONLY. “What about the kids?” “They’ll be blindfolded for pin-the-tail-on-the-donkey.” “That’s not a thing anymore. They’ll be on tablets.” “Then we bring our own donkey. Bill, focus. Phase one: I distract the shark teenager with a complaint about ball-pit hygiene. Phase two: you slip into the party room and replace the presents with empty boxes.” Bill rubbed his temples and looked at the ceiling of the van. “Empty boxes we brought?” “Empty boxes we brought, yes. It’s called a switcheroo.” “It’s called a felony.” “It’s called a birthday felony. There’s a difference.” Molly folded the map into a tiny pirate hat and placed it on the dashboard. Bill stared at the invitation, then at the empty sacks in the back of the van. Molly tapped the dashboard twice, her expression shifting from cheerful to dangerously practical. Bill sighed, folded the napkin, and tucked it into his pocket like a warranty he already regretted. Outside, a delivery van rumbled past, and Bill glanced at the empty sacks in the back, then at the invitation, then at Molly, who was already humming a pirate shanty and checking the van’s glove box for a fake mustache, a roll of wrapping paper, and the sock puppet she insisted was essential to any civilized felony. She had also brought a clipboard, three decoy gifts, and a whistle that she claimed could summon a conga line in under six seconds. Bill prayed silently for the clown to retire early.
created_at
"2026-10-20T08:17:52.618Z"
10/20/2026, 8:17:52 AM
model
"deepseek/deepseek-v4.1-flash-20261017"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
144
moderation_latency
(null)
generation_time
56359
tokens_prompt
75
tokens_completion
9332
native_tokens_prompt
67
native_tokens_completion
10078
native_tokens_completion_images
(null)
native_tokens_reasoning
9403
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
0.00504436
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1792484272-LkCmMbJay5A8tfbUzHDE"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1792484272-ccMicCULqiSr1IHcgZ2x"
upstream_id
"gv2_tGZ996HH63W50LbqzpXUj"
provider_responses
0
endpoint_id
"4231aea9-73af-4a2f-bd14-660b1d8b13b2"
id
"gv2_tGZ996HH63W50LbqzpXUj"
is_byok
false
latency
144
model_permaslug
"deepseek/deepseek-v4.1-flash-20261017"
provider_name
"InferenceNet"
status
200
total_cost
0.00504436
cache_discount
(null)
upstream_inference_cost
0
provider_name
"InferenceNet"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Matches word count
n/a
neededClean
false
words
500
96.99%
Dialogue to Total Word Ratio
Ratio: 50.78%, Deviation: 0.78%
neededClean
false
wordsTotal
512
wordsDialogue
260
98.4972%