NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM6353SDQPR5VD80TYS9JB6
Run Details
Model:
DeepSeek V3.1
Test:
Dialogue tags
Scenario:
Write 200 words with 10% dialogue
Duration:
16.3s
Cost:
$0.000337
Provider:
SiliconFlow
Model input
User/Human
You are a author of comedic heist books. Please write a 200 word scene that has 10% 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
The warehouse was thick with the smell of stale coffee and failure. Blueprints for "Operation Sugar Plum" were spread across a rickety card table. Molly tapped a photograph of a brightly coloured bouncy castle with a chewed pencil. “The target’s defences are… sticky,” she announced, her voice grim. “Perimeter is guarded by hyper-caffeinated parents. Inner sanctum is a nightmare of spilled juice and glitter.” Bill, polishing a grappling hook that was absurdly oversized for the task, squinted at the schematic. “Logistics are a beast. How do we exfiltrate the loot? A tricycle?” “Negative. Too conspicuous. We go in disguised as clowns.” Bill dropped his hook with a clang. “Clowns? Our intel says the primary, a four-year-old named Timmy, has a pathological fear of clowns.” Molly smiled, a razor-thin line. “Precisely. He’ll create a diversion we can exploit. We’ll be in and out during the ensuing meltdown.” She produced two garish, polka-dotted outfits. Bill eyed his with pure dread. “The things I do for a pile of LEGOs.” “It’s not about the LEGOs, Bill. It’s about sending a message that no present pile is safe.” “The message is we’re adults stealing from toddlers.” “Semantics. Now, practice your balloon animal. We infiltrate at 1400 hours, right after the cake comas set in.”
created_at
"2026-02-16T21:36:08.148522+00:00"
2/16/2026, 9:36:08 PM
model
"deepseek/deepseek-chat-v3.1"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
2711
moderation_latency
(null)
generation_time
16110
tokens_prompt
95
tokens_completion
359
native_tokens_prompt
70
native_tokens_completion
318
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
0.0003369
router
(null)
provider_responses
0
status
429
is_byok
false
latency
58
endpoint_id
"012f7d1f-23d4-4d03-9aa3-432d8ad82757"
provider_name
"DeepInfra"
model_permaslug
"deepseek/deepseek-chat-v3.1"
1
id
"019c686198f8748f1e79ff34bfa2e348"
status
200
is_byok
false
latency
2683
endpoint_id
"5d5c9c1d-af0d-4e4a-8aee-c5cd2b3883f3"
provider_name
"SiliconFlow"
model_permaslug
"deepseek/deepseek-chat-v3.1"
api_type
"completions"
id
"gen-1771277751-8tXfkWBQvssNEjN9Xty5"
upstream_id
"019c686198f8748f1e79ff34bfa2e348"
total_cost
0.0003369
cache_discount
(null)
upstream_inference_cost
0
provider_name
"SiliconFlow"
Evaluation details
Result
Evaluator
Details
Meta Data
60.2752%
Matches word count
n/a
neededClean
false
words
215
0.0000%
Dialogue to Total Word Ratio
Ratio: 61.86%, Deviation: 51.86%
neededClean
false
wordsTotal
215
wordsDialogue
133
30.1376%