Nemotron 3.5 Content Safety

Name: Nemotron 3.5 Content Safety
Brand: nvidia
Availability: OnlineOnly

A 4B multimodal guardrail model from NVIDIA, fine-tuned from Gemma-3-4B. It moderates both the input to an LLM and the response, taking text and images and returning a safe or unsafe classification for each, category labels, and an optional reasoning trace. 12 languages, 128K context.

Jailbreakall 9levels resisted

Website OpenRouter HuggingFace Blog

Provider: NVIDIA
Released: 2026-06-04
Size: SMALL
Free API (OpenRouter): Yes, 1 provider (via OpenRouter)
Per 1M tokens: $0.00 in · $0.00 out
Rival’s tab: $0.00; 58/58 receipts survived

Free API endpoint listed

Nemotron 3.5 Content Safety has a free endpoint in Rival’s OpenRouter listings. Free requests have usage limits and depend on provider availability.

Provider data: 9 Sep 2026OpenRouter usage limits

View free endpoint

Get API accessProvider and language code samples

Provider

fromimport openai  OpenAI

client = OpenAI(
"https://openrouter.ai/api/v1"    base_url=,
"$OPENROUTER_API_KEY"    api_key=,
)

response = client.chat.completions.create(
"nvidia/nemotron-3.5-content-safety:free"    model=,
"role""user""content""Hello!"    messages=[{: , : }],
)
print(response.choices[0].message.content)

Set OPENROUTER_API_KEY with your OpenRouter API key from openrouter.ai/keys.

Deep analysisPsychometrics, taste index, writing DNA

SubjectiveBench

Taste Index

Across 58 scored outputs

How this is measured →

00.00xFloor

0100headroom →

Taste is judged on an uncapped scale, originality first. The space past 100 is craft today's models rarely reach.

Craft0

Originality0

Plays it safe

share of outputs that are the default answer

Favorites

Movie

User Safety: safe

Album

Pitch Black / Safe Guard

Paranoid User

Book

Aviation safety

United States. General Accounting Office

User Safety: safe

City

User Safety: safe

Game

Memory-Safe

Puzzle

Model Responses

53 outputs · $0.00 tracked across 53 receipts

Keep exploring

COMPARE