> For the complete documentation index, see [llms.txt](https://docs.sea-lion.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sea-lion.ai/models/sea-lion-guard-v2.md).

# SEA-LION-GUARD v2 (Latest)

**SEA-LION-Guard-v2** is our next-generation suite of action governance and safety models tailored specifically for the Southeast Asian AI ecosystem. Moving beyond standard binary content filtering, the v2 suite introduces real-time **Action Governance**, **Bring Your Own Policy (BYOP)** enforcement, and deep **Multi-Turn Context Evaluation** across 8 core regional languages (Burmese, English, Filipino, Indonesian, Malay, Tamil, Thai, and Vietnamese).

## The Gemma-SEA-LION-Guard-v2 Collection

The v2 collection offers scalable parameters and context windows designed for diverse deployment requirements—from edge agent execution to enterprise cloud governance:

* [**Gemma-SEA-LION-Guard-31B/E4B/E2B-v2**](/models/sea-lion-guard-v2/gemma-sea-lion-guard-v2.md) (`https://huggingface.co/collections/aisingapore/sea-lion-guard-v2`) is a lightweight **31B, E4B and E2B parameter model** with **256K context window** for 31B and **128K context window** for E4B and E2B. Optimized for ultra-low latency, it acts as an efficient edge guardrail for real-time tool interception and rapid API checks.

## Key Enhancements in v2

Compared to legacy moderation models, Gemma-SEA-LION-Guard-v2 shifts from freeform string generation to type-safe, calibrated decision probabilities across three fields:

1. **`content_moderation`**: Categorical probability distribution across `"safe"`, `"sensitive"`, and `"unsafe"` tiers (summing to 1.0).
2. **`rule_applicability`**: Predicate relevance score ($0.0 - 1.0$) indicating whether a custom BYOP policy applies to the interaction.
3. **`rule_violation`**: Calibrated probability score ($0.0 - 1.0$) quantifying the likelihood of a policy breach to enable real-time tool call abortion before execution.

## Our Commitment to Safe Regional AI

Gemma-SEA-LION-Guard-v2 empowers developers, enterprises, and public sector organizations across Southeast Asia to deploy autonomous agents with confidence. By providing structured, calibrated decision outputs, these models ensure AI systems remain culturally aligned, policy-compliant, and secure against emerging agentic risks.

> **Note:** For integration examples using OpenAI-compatible Decisions APIs, local `vllm` deployment scripts, and benchmark results on **SEA-SafeguardBench**, please refer to the individual model documentation pages linked above.


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.sea-lion.ai/models/sea-lion-guard-v2.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.