ngrok Release notes and new features

Create embeddings for search and retrieval through AI Gateway

Turn text into vectors with OpenAI, Gemini, Qwen, Tencent, and Mistral embedding models, using your ngrok access key and credits.

Add ngrok to Next.js, FastAPI, Axum, and Gin

Install an official ngrok package for Next.js, FastAPI, Axum, or Gin to give your app a public ngrok endpoint when it starts.

Use Gemini Interactions through AI Gateway

Send Gemini Interactions API requests through AI Gateway with Google's Gen AI SDK for Python or TypeScript.

GPT-6.1 Sol is now available in AI Gateway

Call OpenAI's GPT-6.1 Sol through AI Gateway with your ngrok access key and credits.

Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna are now available in AI Gateway

Call Anthropic's Claude Opus 5.5 and OpenAI's GPT-6 Sol and GPT-6 Luna through AI Gateway with your ngrok access key and credits.

• Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna are now available in AI Gateway
• xAI's Grok models are now available in AI Gateway

Use GLM-5.3-FlashX and GLM-5.3 through AI Gateway

Try the latest GLM additions with your ngrok credits, without creating separate provider accounts or managing their API keys.

• Use GLM-5.3-FlashX and GLM-5.3 through AI Gateway
• Install ngrok skills from Cursor's plugin marketplace

Add ngrok skills to your coding agent

Install eight new skills that teach Claude, Cursor, and other coding agents to secure endpoints, receive webhooks, test MCP servers, and troubleshoot ngrok.

Jev is now available on ngrok.ai

Call Jev by Typesafe AI on ngrok.ai to get a typed answer and a calibrated probability for each question you send, billed to your ngrok credits, or try it instantly with new playground templates and side-by-side sessions.

Hand endpoint setup to your coding agent

Copy a setup prompt from the dashboard so your coding agent can start ngrok for your endpoint and check the connection.

Test richer prompts in the AI Gateway Playground

Attach images and text files, inspect reasoning and tool calls, and retry individual models while comparing responses in the Playground.