NC Bench
Overview
About
Tests
Models
Vendors
Run: 2026-02-16-01KHM5QH4K6728E89G9BDPPRP5
Run Details
Model:
DeepSeek V3.1
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
45.7s
Cost:
$0.000556
Provider:
SambaNova
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
The attic room above Burford’s Bakery smelled of cinnamon and regret. Molly, her ginger hair escaping a messy bun, was frowning at a schematic spread across a wobbly card table. Bill, a man whose face suggested a bloodhound trying to do long division, was meticulously polishing a pair of jeweller’s loupes. “We’re overcomplicating it,” Molly declared, tapping the schematic. “It’s a child’s birthday party at a petting zoo, not a G8 summit.” Bill didn’t look up. “Chaos is the enemy, Molly. Uncontrolled variables. The zebra, for instance. Nobody expects the zebra.” “It’s a miniature donkey painted with stripes, Bill. It says so on the invoice from ‘Zany Zane’s Zoo-dunit.’ My point is, our point of entry is the problem.” She jabbed a finger at the drawing. “We go in as staff. You’re the new face at ‘Happy Clappy Cakes,’ delivering the six-foot-tall unicorn cupcake. I’m with the balloon-animal-twisting contingent.” Bill finally set his loupes down. “You can’t twist a balloon animal.” “I watched a fifteen-minute tutorial. I can make a profoundly sad-looking worm. It’s all about commitment to the bit. While you’re wheeling in the cake, creating a dessert-based diversion, I case the gift table. The loot is there, just sitting in a glorified picnic area. We’ll be in and out before little Jessica blows out her candles.” “And the exit strategy?” Bill asked, his voice a low rumble. “Assuming we’re not besieged by sugared-up seven-year-olds and a disgruntled striped donkey.” “The septic tank truck,” Molly said with a triumphant grin. “It’s scheduled to pump out the zoo’s porta-loos at 3 PM sharp. We stash the presents in the empty tank compartment. No one, and I mean no one, is going to search a truck full of… well, you know.” Bill stared into the middle distance, contemplating the sheer, audacious stupidity of it. A plan so ridiculous it just might work. “The smell,” he stated flatly. “The presents will smell.” “They’re wrapped, Bill! We’ll double-bag them. It’s a minor olfactory setback for a major financial windfall. We’re talking the latest plastic monstrosities, gift cards, probably a few cheques from generous grandparents. It’s a goldmine.” He sighed, a sound that carried the weight of a thousand bad decisions. “Fine. But if we get caught, we’re not just thieves. We’re the sociopaths who stole from a child’s party and then hid the loot in a sewage truck. Our mugshots will be next to the dictionary definition of ‘despicable.’” “We won’t get caught,” Molly said, rolling up the schematic with a definitive snap. “We’ll be the ghosts of birthday parties past. Now, come on. We need to rehearse your cake-delivery smile. It currently looks like you’re suspecting the cake of arson.” Bill attempted a smile. It was, indeed, a terrifying sight. “Better,” Molly lied. “Just stick to scowling. It’s more authentic. And remember: watch out for the donkey. He’s the wild card.”
created_at
"2026-02-16T21:30:16.616931+00:00"
2/16/2026, 9:30:16 PM
model
"deepseek/deepseek-chat-v3.1"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1199
moderation_latency
(null)
generation_time
45696
tokens_prompt
95
tokens_completion
805
native_tokens_prompt
67
native_tokens_completion
728
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
0.00055605
router
(null)
provider_responses
0
id
"97500c99-869c-4b6f-9d22-50e08cb6b71d"
status
200
is_byok
false
latency
1199
endpoint_id
"1dda4bb9-c626-4816-bb54-99665323fef0"
provider_name
"SambaNova"
model_permaslug
"deepseek/deepseek-chat-v3.1"
api_type
"completions"
id
"gen-1771277370-f6T6aZzsAk7wmi6qgja7"
upstream_id
"97500c99-869c-4b6f-9d22-50e08cb6b71d"
total_cost
0.00055605
cache_discount
(null)
upstream_inference_cost
0
provider_name
"SambaNova"
Evaluation details
Result
Evaluator
Details
Meta Data
97.6276%
Matches word count
n/a
neededClean
false
words
493
0.0000%
Dialogue to Total Word Ratio
Ratio: 70.59%, Deviation: 40.59%
neededClean
false
wordsTotal
493
wordsDialogue
348
48.8138%