Sencha vs Local LLM runner

A native LLM client for on-device inference, LLM APIs, and self-hosted servers, built around a minimal, responsive interface. Download a supported model and chat offline without an API key or account. Or connect OpenRouter, direct providers, and compatible endpoints using your own credentials. On-Device Models • Download models from the built-in catalog, including Gemma 4, Qwen 3.5, Bonsai, and LFM 2.5 • Choose model sizes and quantization options to suit your device • Configure context window, reply length, and image detail where supported • Use streaming, thinking traces, image attachments, and web tools with compatible models • Reduce reprocessing of earlier messages with on-disk prompt caching Model capabilities and performance depend on the selected model and available device memory. Connections and Models • Connect OpenRouter, OpenCode, and direct providers • Add OpenAI- or Anthropic-compatible endpoints, including self-hosted runtimes such as Ollama and LM Studio • Use Chat Completions, Responses API, or Messages API • Choose an on-device or connected model per chat • Browse model catalogs and inspect context size, output limits, pricing, and capabilities where available • Configure reasoning effort, prompt caching, routing, and model tools where supported Web Search and Fetch • Use the included free web search via Markdown.new • Or configure Brave Search, Tavily, Exa, Serper, SerpApi, Parallel, or Keenable • Choose search-depth presets for Exa and Tavily • Fetch pages on device or through Markdown.new or Firecrawl • Configure fetch providers and fallbacks independently Files and Output • Attach photos, documents, and PDFs • Send PDFs natively, extract embedded text, run on-device OCR, or render pages as images • Read responses with formatted Markdown, code, mathematical formulas, and tables • Copy generated tables as a table, Markdown, or a complete image Voice Input • Use Apple Speech for on-device dictation • Download local transcription models • Connect API-based transcription providers The app is free, with no ads or app subscription. On-device models require no provider account or API key. Connected services are subject to their own pricing, limits, and data policies.

Run LLM on your iPhone, 100% local, no network (Network needed once to download a model) Models: • Qwen2.5 0.5B Instruct — ~0.4 GB (4 GB RAM) • Llama 3.2 1B Instruct — ~0.8 GB (4 GB RAM) • Qwen2.5 1.5B Instruct — ~0.9 GB (4 GB RAM) • Llama 3.2 3B Instruct — ~1.9 GB (6 GB RAM) • Qwen3 4B Instruct 2507 — ~2.5 GB (6 GB RAM) • Qwen2.5 7B Instruct — ~4.4 GB (8 GB RAM) Also: • Add your own GGUF (for example, from Hugging Face) • Auto-estimated max context from GPU memory; you can change it Requires iOS 18 or later. Tested on iPhone 11, 14 Pro, 15 Pro with iOS 26. Qwen models © Alibaba Cloud. Licensed under Apache License 2.0. Llama 3.2 is licensed under the Llama 3.2 Community License, Copyright © Meta Platforms, Inc. All Rights Reserved. Built with Llama. On-device inference uses llama.cpp (MIT License).