Support for Venice.ai's API

Currently, Venice.ai (a provider of private and uncensored AI) does not handle thinking streams in a manner compatible with TypingMind. The response format should be compatible with OpenAI, as specified in their API documentation. Qwen 3 4b is the only model that appears to stream the “thought” in appropriately, hiding it within a collapsible box.

It would be great to either receive official support for Venice or, at the very least, support for the manner in which they stream reasoning-model responses.