Problem (one or two sentences)

The current magistral-medium-latest configuration reflects the capabilities from the Magistral Medium 1.1 release in July, 2025 (41k token context window, no image input).

Current magistral-medium-latest default config in Roo Code:

Current magistral-medium-latest capabilities from Mistral documentation:

Context (who is affected and when)

Anyone using Mistral as their API provider and wants to use the latest version of Magistral Medium

Reproduction steps

1. Navigate to provider settings
2. Select Mistral as the API provider
3. Select magistral-medium-latest as the desired model

Expected result

Context window is 128,000 tokens | image input IS supported

Actual result

Context window is 41,000 tokens | image input is NOT supported

Variations tried (optional)

Temporary workaround for OpenAI compatible provider works but it should be supported by default

App Version

v3.28.9

API Provider (optional)

Mistral AI

Model Used (optional)

magistral-medium-latest

Roo Code Task Links (optional)

No response

Relevant logs or errors (optional)