Pangolin 1.22 Tunneled Reverse Proxy Adds AI Gateway

Pangolin, the open-source zero-trust remote access platform, released version 1.22, adding an AI Gateway that acts as an identity-aware proxy between AI clients and model providers, supporting OpenAI, Anthropic, Google Gemini, Amazon Bedrock, Google Vertex AI, Microsoft Foundry, OpenRouter, Vercel AI Gateway, and self-hosted options like Ollama and vLLM. The release also moves browser-based SSH, RDP, and VNC resources, private SSH/HTTPS resources, and certificate status information from Enterprise-only to the free Community Edition.

Pangolin, the open-source zero-trust remote access platform, has released version 1.22, introducing a new AI Gateway to manage access to cloud and self-hosted AI models and moving several previously Enterprise-only remote access features into the free Community Edition. The headline addition is the new AI Gateway, a dedicated Pangolin resource acting as an identity-aware proxy between AI clients and model providers. Instead of configuring applications and coding agents to communicate directly with services like OpenAI, Anthropic, or Gemini, administrators can point them to a Pangolin URL and manage access centrally. The gateway works with both commercial cloud services and locally hosted models. Pangolin provides support for OpenAI, Anthropic, Google Gemini, Amazon Bedrock, Google Vertex AI, Microsoft Foundry, OpenRouter, and Vercel AI Gateway. Self-hosters can also connect custom providers such as Ollama, vLLM, Bifrost, and other compatible endpoints running on their own infrastructure. Multiple providers can be attached to the same gateway. This means a single Pangolin endpoint can expose Claude from Anthropic, GPT models from OpenAI, and a locally hosted Llama model running through Ollama. Administrators can use allow and block lists to control which models are exposed through each gateway. AI Gateways can be configured as public or private resources. A public gateway receives a fully qualified domain name and uses Pangolin virtual API keys for authentication. Clients send the virtual key to Pangolin, which verifies the user’s identity and permissions before forwarding the request with the actual upstream provider credentials. Private AI Gateways take a different approach. They are accessible only from devices connected through the Pangolin client, with the authenticated client session serving as the credential. Users do not need a separate Pangolin API key to access the gateway. Beyond the new AI functionality, Pangolin 1.22 adds browser-based SSH, RDP, and VNC resources, originally introduced as Enterprise features in Pangolin 1.19 https://linuxiac.com/pangolin-1-19-tunneled-reverse-proxy-lands-with-built-in-browser-remote-access/ and now available to Community Edition users as public resources. After authenticating through Pangolin, users can start an interactive SSH terminal in their browser, access a Windows machine through RDP, or connect to a remote desktop using VNC. Pangolin’s browser-based RDP also supports clipboard functionality and file transfers. The resources use the same users, roles, single sign-on, and access policies as other Pangolin resources, providing centralized access management instead of requiring each remote service to be exposed and secured independently. Private SSH and HTTPS resources are now part of Community Edition. With the Pangolin client connected, users can SSH to internal systems over the tunnel using pangolin ssh or access internal web applications through HTTPS hostnames available only on the private Pangolin network. Moreover, Community Edition users gain another previously restricted capability in version 1.22: certificate status information for public resources. The interface now shows whether the TLS certificate associated with a resource is valid directly on its resource page. Several smaller improvements are included as well. Administrators can now send test notifications while configuring alert actions to verify email and webhook integrations before an actual alert occurs. Resource labels can be defined through YAML blueprints or container labels. Fixes address log-retention settings and refresh behavior in log tables. The release also includes interface improvements and general bug fixes. For more details, see the announcement https://pangolin.net/news/1-22-release or refer to the changelog https://github.com/fosrl/pangolin/releases/tag/1.22.0 . Image credits: Pangolin