2.2.26

• Added separators for date intervals in the context list.
• Added a new option: Context -> Show context groups on top of the context list, default: False.
• Added a new option: Context -> Show date separators on the context list, default: True.
• Added a new option: Context -> Show date separators in groups on the context list, default: True.

2.2.25

• Improved native API functions handling.
• Improved expert prompts.
• Fixed/improved inline plugin commands handling.

2.2.24

• Added a STOP button that allows stopping indexing at any time.
• Improved conversion of ENUM types when generating a native list of functions for the API.
• Disabled adding functions to the list when the option to execute commands is unchecked.

2.2.23

• Improved context summarization in Llama-index provider.

2.2.22

• Added new API models for "Chat with Files": Google Gemini and Anthropic Claude (see the new section in the documentation: Managing models -> Using other models (non-GPT) for more information on how to set API keys)
• Added new local models for "Chat with Files": Llama 3.1, Mistral, Codellama, and Llama2-Uncensored (requires Ollama installed on the local machine - see the new section in the documentation: Managing models -> Using other models (non-GPT) for more information on how to enable local model support).
• Added abilitify for context summarization using Llama-index models (local models can be used to summarize context titles now).
• Added a new option in Config -> Indexes (llama-index) -> Indexing: Stop indexing on error. If enabled, indexing will stop whenever an error occurs.

2.2.21

2.2.21 (2024-08-20)

• Added inline execution for external function calls.
• Added a stop to file indexing if an error occurs.

2.2.20

• Llama-index upgraded to 0.10.65
• Langchain upgraded to 0.2.14
• OpenAI API upgraded to 1.41.0
• Code syntax highlight changed from Pygments to highlight.js (better performance)
• Added new model: gpt-4o-mini
• Added native API function calls by default
• Added functions importer (from plugins) into assistants
• Fixed: input message disappear on command response in stream mode
• Fixed: assistants tool outputs submitting

2.2.19

• Added a newest multimodal model: GPT-4 "omni": https://openai.com/index/hello-gpt-4o

2.2.18

• Fix: prevent crash if no audio to play.

2.2.17

• Fix: Added prevent try to play audio if empty output.
• Disabled playing finish event on audio or voice control enabled.