NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-10-13-01M4TBWKGV3YB9XGBCV9Q2VH2Z
Run Details
Model:
Ember-1 (Reasoning, Max)
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
2m 39s
Cost:
$0.131852
Provider:
Fireworks
Evaluation summary
Score
Evaluator
Details Link
100.00%
Matches word count
(details)
86.60%
Dialogue to Total Word Ratio
(details)
93.2972%
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
They met, as always, in the parking lot of a shuttered Wendy's, which Bill considered neutral territory. They had been banned from the functioning Wendy's. He unrolled the blueprints across the hood of Molly's car and pinned one corner down with a bag of cheese puffs. "This is it," he said. "The big one." Molly squinted at the paper. "Bill, this is a crayon drawing of a house." "It's a schematic. I had the inside man draft it." "The inside man is six." "He's very thorough for six. Look, he labeled the bouncy castle." The roof was purple. The chimney appeared to be on fire, though Bill insisted that was a flag. Molly rubbed her temples. "Run the job by me one more time. Slowly." Bill produced a laser pointer from 2009 and waved it over the drawing with tremendous ceremony. "Tomorrow, eleven a.m., Timmy Henderson turns seven. Forty presents, one folding table, in the backyard." "We're robbing a child's birthday party." "There's a Nintendo Switch in there, Molly. Timmy told me himself. Twice. The kid could not stop bragging." Molly said nothing, which was her way of saying several things. She had followed Bill into eleven jobs. Three had worked. One had technically worked. The rest were discussed only in court. "Phase one," Bill said, tapping the schematic, "you go in as the balloon animal lady. I already booked you." "I don't know how to make balloon animals." "You've got fourteen hours and the internet. Phase two: I come over the fence dressed as the clown from the party store." "You got banned from the party store." "Which is why I wear the disguise to buy the disguise. It's foolproof." "What's phase three?" "We load the presents into a Hefty bag during the piƱata. Everyone watches the candy. Nobody watches the gift table." The getaway vehicle, Bill explained, was his own car, parked two houses down, complete with a car seat he described as versatile seating. Molly raised the issue of the hired magician. Bill dismissed it on legal grounds he declined to cite, having read them somewhere. "What if somebody stops us?" she asked. "Nobody stops a clown and a balloon lady carrying a garbage bag. We're invisible. We're party infrastructure." Bill folded the blueprint with the gravity of a man who had never once been right. "Forty presents. Zero security, unless you count Timmy's grandma, and she falls asleep after cake." "Everyone falls asleep after cake. That's not intelligence, that's biology." "So you're in?" Molly sighed the sigh of a woman whose other option was honest work. "I'm in. But I'm only learning one balloon animal. Everything is a snake. Dogs, swords, hats. Snake." "Deal. Nobody inspects the snake." "And if one kid cries, we walk. We're thieves, not monsters." They shook hands over the hood, two professionals with a crayon drawing and a plan so flawed it could only be foiled by a magician. Neither went home to practice balloon animals. One of them really should have.
created_at
"2026-10-13T10:10:49.511Z"
10/13/2026, 10:10:49 AM
model
"fireworks/ember-1-20260923"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1779
moderation_latency
(null)
generation_time
158533
tokens_prompt
75
tokens_completion
8357
native_tokens_prompt
149
native_tokens_completion
8787
native_tokens_completion_images
(null)
native_tokens_reasoning
8120
native_tokens_cached
148
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
0.1318524
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.21.0; linux; x64))"
http_referer
(null)
request_id
"req-1791886249-2u7YxxgNXfXwXdekULRs"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1791886249-wxsUELzaR9ywUZr3rWc7"
upstream_id
"chatcmpl-80dad4e9505240e584fbb3d5a48a0027"
provider_responses
0
endpoint_id
"b89a2f05-4051-4446-a9b0-7bc8ae7b7c90"
id
"chatcmpl-80dad4e9505240e584fbb3d5a48a0027"
is_byok
false
latency
1779
model_permaslug
"fireworks/ember-1-20260923"
provider_name
"Fireworks"
status
200
total_cost
0.1318524
cache_discount
0.0003996
upstream_inference_cost
0
provider_name
"Fireworks"
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
501
86.60%
Dialogue to Total Word Ratio
Ratio: 51.70%, Deviation: 1.70%
neededClean
false
wordsTotal
501
wordsDialogue
259
93.2972%