Ollama Context Viewing / Manual Settings

[RufiS]

It would be nice to display the maximum context length accurately instead of just displaying "1" for Ollama provider. Since the existing Roo-Code code already determines the context length of the chosen Ollama model anyways. Also with that said, the new Qwen3-Coder supports 256k, but I'm forced to run (with 24GB VRAM) at 170k. It would be nice to have the ability to choose the context window size when selecting Ollama as the provider so you can use less than the Ollama model says it supports when retrieving that information from the Ollama server.