NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-05-10-01KR8XQJXCAG1XWJ49WHKQWEW3
Run Details
Model:
Gemma 4 31B
Test:
Dialogue tags
Scenario:
Write 500 words with 30% dialogue
Duration:
12.8s
Cost:
$0.000229
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0000%
Overall
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
Molly spread a crayon-drawn map across the coffee table, weighting the corners down with half-eaten bags of cheese puffs. She pointed a glittery pointer at a smudge of blue ink representing the backyard patio. Bill stared at the map with the intensity of a man trying to solve a Rubik's cube in a windstorm, his brow furrowed in deep professional concentration. "The timing is everything, Bill," Molly whispered, her voice trembling with artificial urgency. "The cake is served at two. That is the window. Total chaos. Sugar crashes. Absolute anarchy." Bill scratched his chin, looking skeptical. "What about the perimeter? I heard the host hired a professional clown. Those guys are unpredictable." Molly rolled her eyes so hard she nearly lost her balance. "He is a balloon artist, Bill, not a Navy SEAL. We slip through the hedge, navigate the inflatable bounce house, and hit the gift table before the piƱata drops." Bill leaned in, his eyes narrowing. "And the security?" "A golden retriever with a penchant for tennis balls," Molly replied. "I have a pocket full of premium treats. He will be our unwitting accomplice." Bill paced the room, his oversized suit jacket flapping. He was a man who took theft seriously, even when the target was a collection of Lego sets and glittery dollhouses. He stopped abruptly, pointing a finger at the map. "The bounce house is a bottleneck," Bill argued. "If we get stuck in the mesh, we are sitting ducks for a mob of seven-year-olds. We need a diversion." Molly smirked, reaching into her bag to produce a small, handheld bubble machine. "I have already accounted for that. I trigger the bubbles on the north lawn, the children swarm the iridescent spheres, and we vanish into the shadows of the gazebo." Bill looked at the bubble machine, then back at Molly. A slow, appreciative grin spread across his face. "You are a monster, Molly. A complete professional." "I just know how children work," she said, checking her watch. "They are easily swayed by soap and water." Bill straightened his tie, feeling the adrenaline surge. "What's the exit strategy? We cannot exactly walk out the front door carrying twelve oversized boxes of Barbie Dreamhouses." Molly tapped the map, specifically a large, plastic laundry hamper hidden behind a hydrangea bush. "The hamper method. We load the loot, slide it across the grass, and meet at the getaway tricycle." Bill paused, his expression falling. "Wait, we are using a tricycle?" "It is the only vehicle that blends in," Molly snapped. "Now, put on your oversized party hat. We have a schedule to keep."
created_at
"2026-05-10T12:27:57.749Z"
5/10/2026, 12:27:57 PM
model
"google/gemma-4-31b-it-20260402"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
826
moderation_latency
(null)
generation_time
12564
tokens_prompt
77
tokens_completion
673
native_tokens_prompt
79
native_tokens_completion
576
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)
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
0.00022915
router
(null)
provider_responses
0
endpoint_id
"4149ab03-3b69-45a0-b94c-6c253b8641a1"
id
"chatcmpl-RGd0J2EVR0jBGPK1dYfy1URS"
is_byok
false
latency
58
model_permaslug
"google/gemma-4-31b-it-20260402"
provider_name
"DeepInfra"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1778416077-KMeBDM03t8a0EjzaGVEV"
session_id
(null)
api_type
"completions"
id
"gen-1778416077-LH2o3OKknGhroSB4AinO"
upstream_id
"chatcmpl-RGd0J2EVR0jBGPK1dYfy1URS"
total_cost
0.00022915
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
434
0.00%
Dialogue to Total Word Ratio
Ratio: 53.65%, Deviation: 23.65%
neededClean
false
wordsTotal
438
wordsDialogue
235
0.0000%