SpeakGPT

SpeakGPT is an advanced and highly intuitive AI open-source chatbot and voice assistant based on different large language models to provide you with unparalleled performance and functionality. This state-of-the-art voice assistant is designed to cater to your every need, whether you require assistance with writing creative stories, learning new and interesting facts, or performing device routines. Officially SpeakGPT supports GPT models, LLAMA, MIXTRAL, GEMMA, Gemini (regular and pro), Vision, DALL-E and more. You can use others models and API endpoints and provide your feedback.

WARNING: This app uses API access. Some API provides use pay-as-you-go subscription. You will need to provide your API key to use this app.

If you chose OpenAI as your API provider:
You can revoke your API key anytime at: https://platform.openai.com/api-keys

Learn about pricing here: https://openai.com/pricing

Information for users who want to use Google Gemini API with this app.

SpeakGPT does not support Google API keys itself, but you cen still use Google Gemini using OpenRouter API. More info at [OpenRouter Models](https://openrouter.ai/docs#models).

For more providers please refer it's documentation, Terms of Service and Privacy Policy.

Features list:
● Chat (chat history is saved locally and can be imported/exported is needed)
● Images generation (DALL-e)
● GPT 4 Vision access (use your photos and images with GPT)
● Custom AI models (fine-tuned) are supported
● Activation prompt
● System prompt
● Assistant
● SpeakGPT in context menu
● SpeakGPT in Share sheet
● Prompts store
● A lot of different models (GPT 4 is supported!)
● No annoying captchas and paywalls
● Pay as you go system
● Tips for newbies
● Advanced support
● Different API providers
● Set LLM params like temperature, top_p, frequencyPenalty, presencePenalty and logit_bias
● Compatible with OpenRouter

This app is open-source. Report bugs, ask a questions and view code here: https://github.com/AndraxDev/speak-gpt

Recent changes:
- Added ability to share images and text messages
- Fixed amoled theme
- /imagine command is now enabled by default*
- gpt-4o model is set as default model.*
- DALL-e 3 is now default model for image generation. Default resolution for generated images updated to 1024x1024 because it now only resolution supported by DALL-e 3*
- Model names in chat list are hidden by default to simplify user interface

* These default settings applies only for new users and chats.