VOLlama v0.7.0

Change log

• Now you can specify the name of embedding model (default EmbeddingGemma) in the rag settings. Embedding only works for OLlama.
• You can now mute the sound effect.
• When launching for the first time, it'll display the dialog for API Settings.
• It now displays friendlier message when unable to retrieve model list and displays API settings dialog.
• Resetting the setting quits the app.
• Both Windows and MacOS Now use Python 3.13.7.

Download

VOLlama v0.6.0

Changelog

• Windows only: Added option to toggle screen reader in the chat menu.
• Mac: Switched to NSSpeechSynthesizer.
• Mac: Organized voices into submenus.
• You can now use the default voice set in System Settings > Accessibility > Spoken Content.
• Added ability to reset settings from the Advanced menu.
• Added warning when resetting settings that are not compatible with the current version.
• Startup speech now triggers only when Read Response is checked.

VOLlama v0.5.0

Change log

• Upgraded dependencies to support GPT-5
• You can specify a direct URL to an image when using the attach URL feature.
• The prompt field now accepts both Shift+Enter and Ctrl+Enter for a new line.
• Changed "Speak response" to "Read response" in the chat menu, so it now uses "R" as the accelerator on Windows.
• Typo fix

Download

VOLlama v0.4.0

Change log

• You can now attach multiple images at once for vision language models that support multiple images such as Gemma3, Qwen2.5VL, Llama Vision.

Download

VOLlama v0.3.2

Changelog

• Displays presets in alphabetical order.
• Removed deprecated parameters for Ollama: penalize_newline, mirostat, mirostat_tau, mirostat_eta, tfs_z
• When temperature is not set, llama-index-llms-ollama sends none as temperature value in order to use the default temperature set by the model.

Download

VOLlama v0.3.1

ChangeLog

• Fixed the bug where selecting a preset doesn't set the system message.
• Selecting a preset starts a new chat.

Download

VOLlama v0.3.0

Changelog

• Use Chat > Presets to save server type, model, system message, and generation parameters.
• Press Ctrl/Cmd+p to quickly switch between your favorite presets.
• Settings aren’t saved to a preset until you click Save in the Preset menu.

Download

VOLlama v0.2.2

Change log

• Correctly map num_predict to max_tokens.
• Important: Set num_predict to a positive value, such as 1024.
• Parameters (except num_ctx) can be left empty to use the engine’s default values.
• The system message can be empty to use the model’s default.
• You can now specify the num_gpu parameter to set the number of layers loaded to the GPU for Ollama.

Download

VOLlama v0.2.1

Change log

v0.2.1

• No longer throws an error when there's no settings for new users.
• Fixed the bug for follow up question with image
• Image, document, and URL uploaded mid generation don't get cleared.
• Only clear prompt when you press escape and you were focused on a previous message.
• List all models returned from server
• Upgrade dependencies
• Other minor bug fixes and performance improvements

Download

VOLlama v0.2.1-beta.2

Changelog

• Fixed a bug for follow up question with image
• Image, document, and URL uploaded mid generation don't get cleared.
• Only clear prompt when you press escape and you were focused on a previous message.
• List all models returned from server
• Upgrade dependencies
• Other minor bug fixes and performance improvements

Download