NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-09-03-01M1KBYZW88K5K0HSMREHPA5ZJ
Run Details
Model:
DeepSeek V4 Pro 0813 (Reasoning, High)
Test:
Language Writing
Scenario:
Character dialogue (Spanish) in a story
Duration:
1m 4s
Cost:
$0.018516
Provider:
Phala
Evaluation summary
Score
Evaluator
Details Link
100.00%
Parse dialogue
(details)
Model input
System
You are an expert novel author in the cozy mystery genre. Take the following details about characters/locations/... into account: <character name="Maria"> Maria is a 30-year old college student on a study abroad program. Born in Spain, she is fluent in Spanish and does not yet know any English. She enjoys landscape photography and is always seen with her camera. </character> <character name="Hunter"> Hunter is a 35-year old detective who is investigating a series of mysterious disappearances in the small town of Alborn, MN. He is a former marine and has a no-nonsense attitude. Hunter has no knowledge of Spanish, not even basic phrases. </character> <location name="Alborn, MN"> Alborne is a small town in the Midwest with a population of 3,000. It is known for its annual apple festival. </location>
User/Human
Write the following scene: Maria is sitting at a cafe, looking at her camera. Hunter approaches her and asks her about a recent event in the town. She responds troubled and shaken in Spanish, and Hunter looks confused. They talk for a while, until Hunter gets a call and has to leave. Return the dialogue between Maria and Hunter as a voice sheet: Maria: "..." Hunter: "..."
Model response
Messages
Meta
AI
**Voice Sheet** Maria’s lines are in Spanish; English translations appear in brackets. Maria (sitting at the café, staring at her camera screen, to herself): “No debería haberla tomado.” [I shouldn’t have taken it.] Hunter (approaching): “Maria? I’m Detective Hunter. I need to ask you about last night, near the old cider mill.” Maria (startled, clutching the camera): “¿Cómo sabe mi nombre? No he hecho nada malo.” [How do you know my name? I haven’t done anything wrong.] Hunter: “Ma’am, I don’t understand. Do you speak English?” Maria (shaking her head): “No hablo inglés. Solo español.” [I don’t speak English. Only Spanish.] Hunter (slower, pointing at the camera): “Okay. Photo? Picture? The man last night—did you see him?” Maria (troubled): “No quiero hablar del hombre. Por favor, no me pregunte más.” [I don’t want to talk about the man. Please don’t ask me anything else.] Hunter: “I don’t know what you’re saying. But you’re scared. That much I can see.” (He points at the camera again.) “Can I see the camera?” Maria (pulling the camera close): “No. No quiero mostrarle mis fotos. No tengo nada que ver con la desaparición.” [No. I don’t want to show you my photos. I have nothing to do with the disappearance.] Hunter (frustrated, rubbing his forehead): “This isn’t working. Do you have a phone? A translator? Somebody who speaks English?” Maria (voice trembling): “Estoy sola aquí. No conozco a nadie. Tengo miedo.” [I’m alone here. I don’t know anyone. I’m scared.] Hunter: “All right, all right. I’m not going to hurt you.” (His phone buzzes. He glances at it.) “Excuse me.” (answering) “Hunter. Yeah. When? … No, I’m with a witness right now. … Fine. I’m on my way.” Hunter (to Maria, urgent): “I have to go. Stay in town. Do not leave Alborn. Do you understand? Stay. Here.” Maria: “No entiendo. ¿Por qué se va? ¿Qué está pasando?” [I don’t understand. Why are you leaving? What is happening?] Hunter (backing away): “I’ll find someone who speaks Spanish. Just stay here.” (He exits.) Maria (to herself, staring at the camera): “Tengo que borrar esta foto.” [I have to delete this photo.]
created_at
"2026-09-03T10:10:21.207Z"
9/3/2026, 10:10:21 AM
model
"deepseek/deepseek-v4-pro-20260813"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
1947
moderation_latency
(null)
generation_time
63314
tokens_prompt
304
tokens_completion
4343
native_tokens_prompt
342
native_tokens_completion
4133
native_tokens_completion_images
(null)
native_tokens_reasoning
3526
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
0.01851578
router
(null)
user_agent
"langchainjs-openai/1.0.0 ((node/v24.19.0; linux; x64))"
http_referer
(null)
request_id
"req-1788430221-8IeO4cdn6q8rvLpmArMG"
session_id
(null)
preset_id
(null)
api_type
"completions"
id
"gen-1788430221-qkho2dWh16JeHA1JLkBp"
upstream_id
"req_c9316769a75c153571816e35b8ffd46e"
provider_responses
0
endpoint_id
"928ab2e9-a85f-427f-adcb-15b2394c70d5"
is_byok
false
latency
279
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"DigitalOcean"
status
429
1
endpoint_id
"bb1fb528-2000-460c-a65e-b82dc347c019"
id
"req_c9316769a75c153571816e35b8ffd46e"
is_byok
false
latency
1947
model_permaslug
"deepseek/deepseek-v4-pro-20260813"
provider_name
"Phala"
status
200
total_cost
0.01851578
cache_discount
(null)
upstream_inference_cost
0
provider_name
"Phala"
response_cache_source_id
(null)
data_region
"global"
workspace_id
"97e315e5-d303-487d-83c1-83180e8a13d4"
Evaluation details
Result
Evaluator
Details
Meta Data
100.00%
Parse dialogue
n/a
dialogue
0
name
"Hunter"
dialogue
"“Ma’am, I don’t understand. Do you speak English?”"
detectedLang
"en"
heavyLang
"en"
scores
en
0.7906698564593302
no
0.5995423340961099
da
0.5939675174013921
tl
0.5752427184466019
sv
0.5460440985732815
cs
0.5333333333333334
pl
0.49784791965566716
hr
0.48604992657856094
sk
0.45567651632970446
de
0.4308943089430894
sq
0.4176372712146422
ro
0.40273037542662116
nl
0.3371212121212121
vi
0.3320610687022901
yo
0.32038834951456313
pt
0.308300395256917
ca
0.2827868852459016
is
0.2813141683778234
sl
0.25531914893617025
it
0.24731182795698922
ms
0.17061611374407584
be
0.16666666666666666
tr
0.16666666666666666
hu
0.1606714628297362
az
0.15254237288135594
et
0.15254237288135594
lt
0.15254237288135594
lv
0.15254237288135594
uk
0.15254237288135594
reliable
true
passes
true
1
name
"Hunter"
dialogue
"“I don’t know what you’re saying. But you’re scared. That much I can see.” “Can I see the camera?”"
detectedLang
"en"
heavyLang
"en"
scores
en
0.8292932741550018
tl
0.576271186440678
it
0.4192799070847852
ro
0.3946731234866829
ca
0.38800489596083226
sv
0.3734335839598998
tr
0.33065595716198126
fr
0.32614555256064687
no
0.32432432432432434
et
0.31787175989085953
da
0.302649930264993
sq
0.29775280898876405
yo
0.2795389048991354
nl
0.24242424242424243
ms
0.23076923076923078
az
0.21875
hr
0.21752738654147105
pl
0.20127795527156547
cs
0.17355371900826447
es
0.1652754590984975
de
0.12587412587412586
sk
0.12587412587412586
be
0.12280701754385964
hu
0.11816578483245149
is
0.11190053285968028
lt
0.11190053285968028
lv
0.11190053285968028
uk
0.11190053285968028
vi
0.11190053285968028
reliable
true
passes
true
2
name
"Hunter"
dialogue
"“All right, all right. I’m not going to hurt you.”"
detectedLang
"en"
heavyLang
"en"
scores
en
0.8275862068965517
tl
0.7255489021956087
no
0.4907407407407407
yo
0.434156378600823
cs
0.4318181818181818
da
0.39293598233995586
de
0.34679334916864607
pl
0.32926829268292684
sk
0.32432432432432434
lv
0.30730478589420657
sl
0.3055555555555556
hr
0.2966751918158568
lt
0.2744063324538259
et
0.24242424242424243
tr
0.22096317280453256
fr
0.21203438395415475
it
0.207492795389049
sq
0.207492795389049
sv
0.207492795389049
reliable
true
passes
true
3
name
"Maria"
dialogue
"“No entiendo. ¿Por qué se va? ¿Qué está pasando?”"
detectedLang
"es"
heavyLang
"es"
scores
es
0.8309859154929576
pt
0.7706422018348624
ca
0.6610169491525424
it
0.635036496350365
fr
0.5548961424332344
ro
0.5222929936305732
sq
0.5106035889070147
tl
0.45750452079566006
hr
0.43820224719101125
lv
0.4371482176360225
cs
0.4263862332695985
en
0.3939393939393939
fi
0.39148073022312374
no
0.3644067796610169
sk
0.358974358974359
pl
0.3273542600896861
sl
0.3055555555555556
lt
0.28741092636579574
yo
0.2628992628992629
ms
0.24433249370277077
eu
0.23857868020304565
da
0.23469387755102045
tr
0.22680412371134018
sv
0.2227979274611399
et
0.20424403183023873
nl
0.20424403183023873
reliable
true
passes
true