Implement the rest of the OpenAI models

[simonw]

Currently just have:

llm-openai-plugin/llm_openai.py

Lines 20 to 25 in e0e8d7d

	"gpt-4o": {"vision": True},
	"gpt-4o-mini": {"vision": True},
	"o3-mini": {"reasoning": True},
	"o1-mini": {"reasoning": True, "schemas": False},
	"o1": {"reasoning": True, "vision": True},
	"o1-pro": {"reasoning": True, "vision": True, "streaming": False},

Need to re-add GPT-4 / GPT-4-turbo and probably a few others too.

[simonw]

llm-openai-plugin/llm_openai.py

Lines 22 to 23 in 807a45b

	"gpt-4.5-preview": {"vision": True},
	"gpt-4.5-preview-2025-02-27": {"vision": True},

I'm not crazy about registering models like that one twice - this means they both get separate model IDs in the logs, even though either of those model IDs uses the same model.

Ideally I'd spot that the models were the same based on data returned by the OpenAI API, but LLM core doesn't have a mechanism yet for models to say "this should be logged as gpt-4.5-preview-2025-02-27 even though the requested model ID was something else".

[mgaitan]

I'm interested in gpt-4o-mini-transcribe and gpt-4o-transcribe