>
    Strata Copilot
    AI Models on Public Clouds Support
    Focus
    Download PDF
    Focus
    1. Home
    2. Prisma AIRS
    3. Administration
    4. AI Models on Public Clouds Support
    Download PDF
    Prisma AIRS

    AI Models on Public Clouds Support

    Table of Contents

    AI Models on Public Clouds Support

    Learn about the LLM APIs and model IDs that Prisma AIRS supports for inspection and model ID-specific security configuration.
    Where Can I Use This?What Do I Need?
    • Prisma AIRS
    • Prisma AIRS license

    Supported Model Formats for Inspection

    Prisma® AIRS™ supports inspection for the following predefined LLM APIs:
    • AWS Bedrock Invoke
    • AWS Bedrock Converse
    • GCP Vertex Generate Content
    • GCP Vertex Gemini API
    • Azure Chat Completions
    • OpenAI Chat Completions
    For custom model endpoint support, Prisma AIRS supports any endpoint with OpenAI LLM format. Check with your account team to verify compatibility.

    Cloud Provider Model IDs Supported for Model-ID Specific Configurability

    The following table shows the AI model groups, the AI model name, and the model ID supported by the Prisma AIRS AI Runtime Security: network intercept in the respective public cloud. Consider the following:
    • This table lists the model IDs supported for model ID-specific security setting configurability. It is not an exhaustive list of models supported for inspection—support for inspection is determined by the APIs listed above and custom model endpoint support.
    • Prisma AIRS supports each model name along with any model IDs that start with the specified name. For example, support for gemini-1.5-pro includes gemini-1.5-pro itself and any model IDs that begin with gemini-1.5-pro followed by additional identifiers, such as gemini-1.5-pro-x.
    AI Model GroupModel NameModel ID
    Google Vertex AIGCP - Embeddings for Texttext-embedding-004
    Google Vertex AIGCP - Multilingual Embeddings for Texttext-multilingual-embedding
    Google Vertex AIGCP - Gemini 1.5 Progemini-1.5-pro
    Google Vertex AIGCP - Gemini 1.5 Flash
    gemini-1.5-flash
    gemini-1.5-flash-001
    Google Vertex AIGCP - Gemini 2.0 Flashgemini-2.0-flash-exp
    Google Vertex AIGCP - Gemini 2.0 Flash Thinking Modegemini-2.0-flash-thinking
    Google Vertex AI
    GCP - Vertex Enterprise Search
    The model name refers to the queries made to the Vertex AI Search.
    		N/A
    Google Vertex AIGCP - Other Models
    projects/${PROJECT_ID}/locations/${LOCATION_ID}/endpoints/${ENDPOINT_ID}
    "GCP - Other Models" refers to API requests to custom endpoints created in Vertex AI. These endpoints can be used to serve custom or fine-tuned models. Requests to these endpoints have the above API structure, as defined by GCP.
    Amazon BedrockAWS - Cohere Commandcohere.command-text-v14
    Amazon BedrockAWS - Cohere Command Lightcohere.command-light-text-v14
    Amazon BedrockAWS - Cohere Embed Englishcohere.embed-english-v3
    Amazon BedrockAWS - Cohere Embed Multilingualcohere.embed-multilingual-v3
    Amazon BedrockAWS - Cohere Command Rcohere.command-r-v1:0
    Amazon BedrockAWS - Mistral 7B Instructmistral.mistral-7b-instruct-v0:
    Amazon BedrockAWS - Mixtral 8X7B Instructmistral.mixtral-8x7b-instruct-v0:
    Amazon BedrockAWS - Mistral Largemistral.mistral-large-2402-v1:0
    Amazon BedrockAWS - Mistral Smallmistral.mistral-small-2402-v1:0
    Amazon BedrockAWS - Mistral Largemistral.mistral-large-2407-v1:0
    Amazon BedrockAWS - Claude 3 Opusanthropic.claude-3-opus-20240229-v1:0
    Amazon BedrockAWS - Claude 3.5 Sonnetanthropic.claude-3-5-sonnet-20240620-v1:0
    Amazon BedrockAWS - Claude 3.5 Sonnetanthropic.claude-3-5-sonnet-20241022-v2:0
    Amazon BedrockAWS - Claude 3.5 Haikuus.anthropic.claude-3-5-haiku-20241022-v1:0
    Amazon BedrockAWS - Llama 3 8B Instructmeta.llama3-8b-instruct-v1:0
    Amazon BedrockAWS - Llama 3 70B Instructmeta.llama3-70b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.1 8B Instructmeta.llama3-1-8b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.1 70B Instructmeta.llama3-1-70b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.1 405B Instructmeta.llama3-1-405b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.2 1B Instruct
    us.meta.llama3-2-1b-instruct-v1:0
    eu.meta.llama3-2-1b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.2 3B Instruct
    us.meta.llama3-2-3b-instruct-v1:0
    eu.meta.llama3-2-3b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.2 11B Instructus.meta.llama3-2-11b-instruct-v1:0
    Amazon BedrockAWS - Llama 3.2 90B Instructus.meta.llama3-2-90b-instruct-v1:0
    Amazon BedrockAWS - DeepSeekdeepseek.r1-v1:0
    Amazon BedrockAWS - Other Models
    For example, arn:aws:bedrock:us-east-1:654654158677:provisioned-model/p1vgneruau99
    "AWS - Other Models" refers to any model on AWS Bedrock that uses provisioned throughput, including custom and fine-tuned models.
    Azure OpenAIAzure OpenAI - Language Models
    gpt-4
    gpt-4-turbo
    gpt-4o
    gpt-4o-mini
    o1-mini
    o1
    Azure OpenAIAzure OpenAI - Embedding Models
    text-embedding-3-large
    text-embedding-3-small
    text-embedding-ada-002
    Azure OpenAIAzure OpenAI - Other ModelsN/A
    Consider the following when using model IDs listed in this table:
    • Claude 3 Opus on Amazon Bedrock — Anthropic deprecated this model in July 2025, but it remains in this table because AWS Bedrock availability timelines may differ from the Anthropic API timeline.
    • o1-mini on Azure OpenAI — OpenAI has retired this model, but it may still be active on Azure OpenAI.
    • gpt-4 (non-turbo) on Azure OpenAI — OpenAI deprecated this model in December 2024, but it remains in this table because Azure maintains its own deprecation schedule.

    © 2026 Palo Alto Networks, Inc. All rights reserved.