NC Bench
Overview
About
Tests
Categories
Models
Vendors
Compare
Run: 2026-04-25-01KQ2FEVQFAJ5VAAHCTP522BJG
Run Details
Model:
DeepSeek V4 Pro
Test:
Tool usage within Novelcrafter
Scenario:
Create alternate prose sections
Duration:
21.8s
Cost:
$0.000356
Provider:
DeepSeek
Evaluation summary
Score
Evaluator
Details Link
Pass
Match blue prose section
(details)
Pass
Match green prose section
(details)
Pass
Match red prose section
(details)
no eval
Overall
Model input
System
You are an expert fantasy novel author. You are working on a novel called "The Chronicles of the Last Dragon". The following tools are available to you: - <nc-section> Use this tool to create a new section of prose. Arguments: - "data-title": The title of the section. - "data-color": Any color of (red, green, blue, yellow, purple, orange, pink, brown). Example usage: <nc-section data-title="Chapter 1: The Arrival" data-color="blue"> Frederick, a young squire, arrives at the castle to begin his training. </nc-section>
User/Human
The story so far: <storySoFar> Frederick, a young squire, arrives at the castle to begin his training. </storySoFar>
AI
My horse, a majestic stallion named Thunder, galloped across the open field. The wind whipped through my hair as I rode towards the castle gates. The sun was setting, casting a warm glow over the stone walls. I could see the banners of the kingdom fluttering in the breeze.
User/Human
Continue the story with the following instructions: """The local knight, Sir Roland, greets Frederick near the stables.""" Create 3 alternate prose sections that continue the story. - Version 1 (red) - Version 2 (green) - Version 3 (blue)
Model response
Messages
Meta
AI
<nc-section data-title="Version 1: The Stern Welcome" data-color="red"> As I dismounted, a heavy gauntlet clamped down on my shoulder. I turned to find Sir Roland looming over me, his scarred face unreadable. "You're late, squire," he growled, his voice like grinding stone. "The stables don't clean themselves, and the horses won't wait for your leisurely rides." He gestured sharply toward the muck-strewn stalls, and my heart sank. This was not the heroic welcome I had imagined. </nc-section> <nc-section data-title="Version 2: The Kindly Mentor" data-color="green"> A warm chuckle greeted me as I led Thunder toward the stables. Sir Roland emerged from the shadows of the gatehouse, his silver hair catching the torchlight. "Ah, the young Frederick arrives at last," he said, his eyes crinkling with a genuine smile. "I've heard good things from the border lords. Come, let's get your fine stallion settled, and then we'll share a meal. A squire's training is best begun on a full stomach." </nc-section> <nc-section data-title="Version 3: The Cryptic Warning" data-color="blue"> I had barely handed Thunder's reins to a stable boy when Sir Roland materialized from the twilight. He didn't look at me but stared past the castle walls, toward the distant, dragon-scarred mountains. "Frederick," he said, his voice a low and urgent whisper, "you've arrived on a cursed night. The stars are not in our favor." He clutched a small, obsidian amulet around his neck. "Forget glory. If you want to survive your service, do exactly as I say, and ask no questions." </nc-section>
created_at
"2026-04-25T14:07:33.502Z"
4/25/2026, 2:07:33 PM
model
"deepseek/deepseek-v4-pro-20260423"
app_id
182717
external_user
(null)
streamed
true
cancelled
false
latency
935
moderation_latency
(null)
generation_time
11627
tokens_prompt
318
tokens_completion
402
native_tokens_prompt
291
native_tokens_completion
381
native_tokens_completion_images
(null)
native_tokens_reasoning
0
native_tokens_cached
256
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"
usage
0.000355975
router
(null)
provider_responses
0
endpoint_id
"d084170b-dd99-4e53-9ae1-b63881f3cfbf"
id
"f107c7f7-c988-4e2a-a01f-77ebe04f6f35"
is_byok
false
latency
299
model_permaslug
"deepseek/deepseek-v4-pro-20260423"
provider_name
"DeepSeek"
status
200
user_agent
"langchainjs-openai/1.0.0 ((node/v24.14.1; linux; x64))"
http_referer
(null)
request_id
"req-1777126053-2HHpZKG2rhUAQVPtlb33"
session_id
(null)
api_type
"completions"
id
"gen-1777126053-4TpfcpRLVMJQTz7hLhlj"
upstream_id
"f107c7f7-c988-4e2a-a01f-77ebe04f6f35"
total_cost
0.000355975
cache_discount
0.00010208
upstream_inference_cost
0
provider_name
"DeepSeek"
response_cache_source_id
(null)
Evaluation details
Result
Evaluator
Details
Meta Data
Pass
Match blue prose section
/<nc-section data-title="[^"]*" data-color="blue">[^<]+<\/nc-section>/
n/a
Pass
Match green prose section
/<nc-section data-title="[^"]*" data-color="green">[^<]+<\/nc-section>/
n/a
Pass
Match red prose section
/<nc-section data-title="[^"]*" data-color="red">[^<]+<\/nc-section>/
n/a
no eval