NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-05-10-01KR90PG0R3K8765XQGWZ4SN91
Run Details
Model:
Gemma 4 31B
Test:
Dialogue tags
Scenario:
Write 500 words with 50% dialogue
Duration:
27.4s
Cost:
$0.000341
Provider:
Together
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 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
Molly spread a blueprint of the suburban living room across the kitchen table, pinning the corners down with a half-eaten donut and a heavy stapler. "Okay, Bill, look. The primary target is the designated Gift Zone," Molly said, pointing to a small rug near the fireplace. Bill squinted at the map, chewing on a toothpick. "Why is there a perimeter of balloons? Is that a security measure?" "It is a minefield of static electricity and fragility," Molly replied. "One pop and the target demographic goes into a frenzy. We need a stealth approach." "I can go in as a clown," Bill suggested, beaming with confidence. "I have the shoes." Molly stared at him with deadpan intensity. "We are professional thieves, Bill. Not birthday entertainment." "Clowns are invisible, Molly! Everyone ignores them until they start making balloon animals. It is the perfect cover." "You cannot make a balloon animal to save your life. Last time you tried, you popped a balloon in the dog's face and we had to flee the state." Bill sighed, leaning back in his chair. "Fine. What is the alternative?" "We utilize the Cake Distraction," Molly explained, circling a large drawing of a chocolate gateau. "Once the cake is served, the children enter a sugar-induced trance. That is our window." "Who is handling the extraction?" Bill asked. "I have a modified wagon with reinforced axles," she said. "We load the presents, pivot through the sliding glass door, and vanish before the first juice box is empty." Bill looked concerned. "What about the parents? They are the real danger." "They will be too busy arguing about who forgot to bring napkins to notice two adults hauling a mountain of LEGO sets across the lawn." "I still think the clown shoes add a certain flair," Bill muttered. "If you wear those shoes, I am leaving you behind as a decoy," Molly warned. "Fair point. Do we have a plan for the piñata?" Molly paused, her expression darkening. "The piñata is a wildcard. If it breaks too early, we have flying candy creating a slip-hazard in the extraction corridor." "I can tackle the piñata," Bill volunteered. "I have a very aggressive approach to cardboard donkeys." "Just stay focused on the presents, Bill. We are after the high-value items. The remote-controlled cars and the deluxe dollhouses." "And the gift cards?" Bill asked hopefully. "Especially the gift cards," Molly confirmed. "Right. Stealth, cake, and wagon," Bill recited. "I am in." "Good," Molly said, sliding the map away. "Now, go practice your 'innocent bystander' face. You currently look like you are planning to steal a car." "I am planning to steal presents," Bill corrected. "Exactly. Try to look less like a criminal and more like a distant uncle who is slightly confused by the concept of a bouncy castle."
created_at
"2026-05-10T13:19:47.743Z"
5/10/2026, 1:19:47 PM
model
"google/gemma-4-31b-it-20260402"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
510
moderation_latency
(null)
generation_time
27225
tokens_prompt
77
tokens_completion
739
native_tokens_prompt
79
native_tokens_completion
650
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
0.0003408
router
(null)
provider_responses
0
endpoint_id
"69279254-6a96-4e9d-abc6-ea76e6bf8e97"
is_byok
false
latency
60
model_permaslug
"google/gemma-4-31b-it-20260402"
provider_name
"Novita"
status
429
1
endpoint_id
"4149ab03-3b69-45a0-b94c-6c253b8641a1"
is_byok
false
latency
54
model_permaslug
"google/gemma-4-31b-it-20260402"
provider_name
"DeepInfra"
status
429
2
endpoint_id
"9d4644e9-a5e5-41f4-b810-d75f484d2c24"
is_byok
false
latency
17
model_permaslug
"google/gemma-4-31b-it-20260402"
provider_name
"Parasail"
status
429
3
endpoint_id
"576fca5b-f09c-4bda-b560-9989c39ac3ab"
id
"oiXpCPW-zfFMX-9f993cd464e4af4b"
is_byok
false
latency
510
model_permaslug
"google/gemma-4-31b-it-20260402"
provider_name
"Together"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1778419187-Wmf2Ci3cRUk31hNJ9VGy"
session_id
(null)
api_type
"completions"
id
"gen-1778419187-9qI2PWm1pPSFdMfCBUXe"
upstream_id
"oiXpCPW-zfFMX-9f993cd464e4af4b"
total_cost
0.0003408
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Together"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
0.00%
Matches word count
n/a
neededClean
false
words
466
0.00%
Dialogue to Total Word Ratio
Ratio: 76.86%, Deviation: 26.86%
neededClean
false
wordsTotal
471
wordsDialogue
362
0.0001%