OpenAI Text and Image Models

Similar to how Venice supports private video models from Google and OpenAi, it should also support private text and image models in both the UI and API and anonymize the call. Leave it to the user / developer to pick the model that makes the most sense for the use-case. To be honest, the private models are leaps and bounds ahead of Qwen3 and the other OSS models, making it hard to justify using Venice given the alternatives are so much better quality.