NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-01-01M1EM21SYB9N6CC2PQ2HWJFVG
Run Details
Model:
Thinking Machines Inkling Small (Reasoning, High)
Test:
Dialogue tags
Scenario:
Write 500 words with 70% dialogue
Duration:
28.1s
Cost:
$0.004123
Provider:
DeepInfra
Evaluation summary
Score
Evaluator
Details Link
0.00%
Matches word count
(details)
0.00%
Dialogue to Total Word Ratio
(details)
0.0001%
Overall
Model input
User/Human
You are a author of comedic heist books. Please write a 500 word scene that has 70% 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 basement smelled like stolen birthday cake and deep regret. Molly adjusted her black turtleneck and pointed at a blueprint taped to the cinderblock wall. “You realize it’s a children’s party,” Bill said, chewing on a cheap cigar. “The security is a six-year-old with a juice box and an overinflated sense of justice.” “Juice box is a weapon, Bill. Read the manual.” “I wrote the manual. Page one: do not rob kids.” “We’re not robbing them. We’re liberating the presents. The parents are already doing drugs, Bill. The economy demands it.” “That is not how revenue works.” Molly rolled her eyes. “Plan A: we enter through the piñata. It’s a structural weakness. Everyone’s distracted when they hit it.” “Plan B: we say we are the entertainment. I can do a magic trick.” “You can’t do magic tricks. You once turned a quarter into a nickel by losing it.” “That requires skill.” “Plan C is the real plan,” Molly said. “We slip through the bounce house. The mesh is porous. We climb the inflatable wall and rappel down with a stolen gift bag.” “I don’t have a stolen gift bag. I have a duffel bag and deep shame.” “We steal the bag at step one.” Bill sighed. “What about the presents themselves? They are wrapped in paper. I can open them?” “No opening until extraction. We load the sled—” “We don’t have a sled.” “We use a stolen tricycle.” “We are adults, Molly.” “We are adult criminals. The tricycle is faster than the parents’ minivan.” Bill pulled out a notepad. “Let’s review. We dress as balloon animals?” “We dress as disappointed relatives. We bring our own sad gift.” “That is psychological warfare.” “Exactly. We blend in, shake hands with the clown, and then—” “You shake hands with the clown?” “Everyone shakes hands with the clown. Then we ask where the gift table is with the authority of a man who has seen things.” “That is you.” “I want to clarify the exit strategy. Do we use the window?” “The window has a sticker that says window. Kids are vigilant.” “We use the cake table ventilation system. I saw a vent behind the frosting.” “That vent is for heat, Molly.” “It is a portal to destiny.” Bill stared at her. “What if we get caught?” “We run. We say we are collecting data for a children’s charity about hope.” “We are not a charity.” “We are a non-profit now.” The clock ticked. Molly pulled out a walkie-talkie. “I’ll wear the earpiece. You wear the fake beard.” “I have a real beard.” “It’s patchy.” “It’s character development.” “We split up at the door. You create a diversion with a fake allergic reaction to peanuts.” “I am allergic to peanuts.” “Perfect. You collapse, they rush to you, I grab the presents.” “What if they call an ambulance?” “We steal the ambulance.” “Where does that fit in the timeline?” “Step five.” Bill rubbed his temples. “The presents are for children, Molly. There is something wrong with this plan.” “There is everything wrong with it. That is why it is art.” “Can we at least pick a target? Like, do we want the LEGO set or the doll?” “We want all of them. We are chaotic neutral.” “I am chaotic good today.” “You are not good. You are just scared.” “I am practical.” “Practical thieves don’t wear black turtlenecks in a jungle gym.” Bill adjusted his collar. “Okay. We enter at seventeen minutes past the hour. The singing of Happy Birthday is at minute thirty. The distraction is maximum.” “Good. I will sing off-key to create confusion.” “You can’t sing.” “I can scream.” “That works.” Molly grabbed her duffel and the notepad. “One last thing. The getaway vehicle is your mother’s Honda.” “She still thinks I am a tax attorney.” “She will find out when she sees us strapped to the roof with a sack of action figures.” Bill sighed. “Let’s do this. But if we steal a teddy bear, I am keeping it.” “Deal. And if they have cupcakes, I want three.” “Two.” “Three.” “Two and a cupcake.” “Fine.” They left the basement and stepped into the afternoon sun, two thieves ready to ruin a childhood and fund a retirement.
created_at
"2026-09-01T13:55:37.929Z"
9/1/2026, 1:55:37 PM
model
"thinkingmachines/inkling-small-20260730"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
599
moderation_latency
(null)
generation_time
28073
tokens_prompt
75
tokens_completion
3563
native_tokens_prompt
74
native_tokens_completion
3408
native_tokens_completion_images
(null)
native_tokens_reasoning
2399
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
0.0041229
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788270937-MiupcQxpT4YNAZKK7XIA"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788270937-jaNfiQpIKsgPZ34npeE3"
upstream_id
"chatcmpl-RuUfYFQOIcPBuegBzLJ5ulHm"
provider_responses
0
endpoint_id
"d1d9657b-3c3a-49b5-8f81-33204e7c71c3"
id
"chatcmpl-RuUfYFQOIcPBuegBzLJ5ulHm"
is_byok
false
latency
406
model_permaslug
"thinkingmachines/inkling-small-20260730"
provider_name
"DeepInfra"
status
200
total_cost
0.0041229
cache_discount
(null)
upstream_inference_cost
0
provider_name
"DeepInfra"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
705
0.00%
Dialogue to Total Word Ratio
Ratio: 86.48%, Deviation: 16.48%
neededClean
false
wordsTotal
710
wordsDialogue
614
0.0001%