AndraxDev/speak-gpt

What it solves

SpeakGPT provides a versatile, open-source client for interacting with various large language models (LLMs) on Android and the web, removing the need to rely on a single provider's official app while allowing users to bring their own API keys.

How it works

It acts as a frontend interface that connects to multiple API providers including OpenAI, GROQ, Azure, and OpenRouter. Users input their own API keys, which are stored locally on the device, to access models such as GPT, LLAMA, MIXTRAL, GEMMA, and Gemini.

Who it’s for

Users who want a customizable AI assistant experience with support for multiple model providers, advanced parameter control, and the ability to use their own API keys for a pay-as-you-go system.

Highlights

• Multi-Model Support: Compatible with GPT, LLAMA, MIXTRAL, GEMMA, Gemini, and DALL-E.
• Multimodal Capabilities: Supports image generation and image recognition.
• Voice Integration: Includes voice input via Whisper and Google.
• Advanced Controls: Allows customization of model parameters like temperature, topP, frequency penalty, and presence penalty.
• System Integration: Features a context menu and share sheet integration for easier access.
• Developer Friendly: Supports custom API providers and fine-tuned models.