Nerdy update alert (2.7.0): Add Anthropic-compatible endpoints, test custom providers in Playground, choose cache duration, and track cache reads and writes

Version 2.6.0 brings Siri AI support, image generation, and a new Playground so you can try ideas before you turn them into Shortcuts. NEW • Siri can use your Projects. On iOS 27, Siri with Apple Intelligence can find unlocked projects and answer from their summaries and files. Locked projects stay private, and you can turn this off in Settings. • Create images from Shortcuts. Choose an OpenAI, Google, or xAI image model in the "Send LLM Request" action, then preview or save the result with the "Get LLM Image" action. • Web search for on-device models. Gemma 4 and Apple Intelligence can look things up on the web, with inline citations. • Playground. Test prompts against any provider and model — including structured JSON — before you add them to a Shortcut. IMPROVED • Attach CSV files to requests, including on-device Gemma 4. • On newer iPhones and Apple Silicon Macs, Gemma 4’s context window doubles to 8K tokens, so longer conversations and attachments go further on your device.

Version 2.3.0 • Removed the Tip Jar. Thank you to everyone who supported the app, it means a lot. LLM API Request remains free and will continue to be maintained and improved. • The app is now available in the EU.

Version 2.2.0 — What's New On-device AI with Gemma 4 Run Google's Gemma 4 models (E2B and E4B) entirely on your device — no API key, no internet connection, and nothing leaves your phone. • Native image understanding: send photos to the model and get answers fully offline. • Download and manage models right inside the app. • Works as a provider in your Shortcuts, just like the cloud providers.

What's New in Version 2.0 New Features * Projects: Organize AI conversations by topic or workflow. Attach files to projects so they remain accessible across sessions without re-uploading. * Apple Intelligence: Utilize Apple’s on-device model directly. No API key is required, and data remains on your device. * Custom Providers: Connect to any OpenAI-compatible LLM endpoint. Ideal for self-hosted models or third-party APIs. * Is Model Connected (Shortcut Action): Check model reachability before running automations to create more reliable workflows. Improvements & Bug Fixes * Inference Timeout: Extended to 5 minutes to support longer responses. * Model Lists: Now filtered to display only chat-capable models. * Stability: General bug fixes and stability improvements.

What's New in 1.1.0 Model selection dropdown — Pick from your provider's available models directly in Shortcuts instead of typing model IDs manually Smarter conversation history — Now includes model and provider info, so downstream Shortcuts actions automatically know what was used Fixed crashes with large images — Switched to memory-efficient image processing to prevent out-of-memory crashes in Shortcuts Reliable token tracking — Token usage now records correctly even when using Shortcuts with the app closed Model caching — Models are cached locally to prevent Shortcuts timeout errors Bug fixes — Improved button tap areas and updated to modern file type APIs