added https://novita.ai/ to api providers

Added Languages

Svenska

English (US)

Deutsch

中文 (简体) (Kinesiska, förenklad)

Русский (Ryska)

Polski (Polska)

Français (Franska)

Español (España) (Spanska, Spanien)

Português (Brasil) (Portugisiska, Brasilien)

Suomi (Finska)

ไทย (Thailändska)

added models to zhipu

added https://cloud.siliconflow.com/

added api chatanywhere (https://api.chatanywhere.tech/v1)

you can now have multiple chat windows

popup.js: I updated the appendMessage function. It now includes logic to add a clickable TTS button to messages with the role 'user', mirroring the functionality that was already present for 'ai' messages.

popup.css: I added a new CSS class (.tts-btn-user) specifically for this new button. Since your user messages are right-aligned, this new style positions the speaker icon on the left side of the message bubble (left: 5px;) to keep it clean.

I have corrected the logic for when the "Image Prompt" section is displayed.

Previously, the "Image Prompt" box was appearing for providers like Gemini, Anthropic, and xAI. This was confusing because while those models support images (they can see and analyze them), they cannot create images through this app's setup.

I've now fixed it so that the "Image Prompt" section only appears when you select an API that can actually generate images:

OpenAI (for DALL-E)

Stable Diffusion (A1111)

This makes the interface cleaner

added negative prompt for automatic 1111

addad to automatic 1111 Number of Steps, CFG Scale, Width, Height, Sampler and seed

added cfg scale for automatic 1111

Added support for image generation with local AUTOMATIC1111

fixed problem that no models was loaded when program started and fixed it so it remembers last api and model

updated models

unsued apis dont longer show in the list

addad so only the apis you have a api kry for is visable

missed context window in ollama model

some bugfixes and have sorted the api in alfabetical order

added on ollama models to change context window size

added text to speach intergration

fixed a bug that google gemini did not analyse screenshots

updated the way the models was loading