OpenGuardrails
OpenGuardrailsAn open, context-aware AI guardrails platform that pairs a safety/manipulation-detection model (OpenGuardrails-Text-2510, a 3.3B quantized LLM) with a deployable guardrails platform: APIs, deployment scripts, and modular components for private or on-prem use. Covers content safety, manipulation attacks (prompt injection, jailbreak, code-interpreter abuse), and data leakage across 119 languages.
OpenGuardrails: Apache-2.0 platform AND model (OpenGuardrails-Text-2510, 3.3B quantized); 119 languages; on-prem/private deployment. ~80 GitHub stars (June 2026); arXiv 2510.19169. The first project to open-source both a large-scale safety LLM and a production guardrails platform. Verified June 2026.
Openness
5 high confidence- license
- Apache-2.0(OSI) for both platform and model
- source
- public
- self-host
- yes(on-prem/private)
- model
- OpenGuardrails-Text-2510 open weights Apache-2.0
- core-gated
- ungated
Fully open source: both the guardrails platform and the safety model are Apache-2.0 and self-hostable, including on-prem deployment.
- https://github.com/openguardrails/openguardrails recorded 2026-06-29
Apache-2.0 platform; ~80 stars; on-prem deployment
- https://huggingface.co/openguardrails/OpenGuardrails-Text-2510 recorded 2026-06-29
Apache-2.0; 3.3B quantized safety model; 119 languages
Adoption
2 low confidenceNew (late 2025) project, ~80 GitHub stars (June 2026); stars-only proxy, capped per methodology; early-stage adoption.
- https://github.com/openguardrails/openguardrails recorded 2026-06-29
~80 stars (June 2026)
Capability
4 low confidenceUnusually complete for its age: ships both an open safety model and an on-prem platform, with broad multilingual coverage.
- https://arxiv.org/abs/2510.19169 recorded 2026-06-29
context-aware guardrails platform + safety model; 119 languages; content safety, manipulation, data leakage
Unchanged since 2026-08-01 (last edited, not re-checked)