[alexcastgonz (pcbeingused333)]

Motivation and Context

MistralAIChatCompletion builds FinishReason(choice.finish_reason) for every choice. FinishReason only defines stop, length, content_filter, tool_calls and function_call, but the Mistral SDK also returns model_length (the model's context window was reached) and error. For those valid responses, both get_chat_message_contents and get_streaming_chat_message_contents raised ValueError: 'model_length' is not a valid FinishReason.

Description

• Added _get_finish_reason in mistral_ai_chat_completion.py: model_length maps to FinishReason.LENGTH, any other value that FinishReason does not define maps to None (logged at debug), and known values behave as before. Both the non-streaming and streaming paths use it.
• Tests (parametrized, non-streaming and streaming): model_length -> LENGTH, an unknown value -> None, length unchanged. They fail on main and pass with the change.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the SK Contribution Guidelines
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone

Disclosure

This change was prepared with AI assistance (Claude Code).

[Copilot]

🟢 Approval recommended

No active review issues are identified for the Mistral AI finish-reason handling update.

No open issues 🎉

What changed in this PR

This PR prevents Semantic Kernel’s Mistral AI chat connector from raising ValueError when responses contain finish reasons outside the shared FinishReason enum.

Changes:

• Maps model_length to FinishReason.LENGTH and returns None for unknown values, with debug logging.
• Uses the mapping for both streaming and non-streaming responses.
• Adds parametrized regression tests for both paths, plus a non-streaming check that length remains unchanged.

File	Description
python/​tests/​unit/​connectors/​ai/​mistral_ai/​services/​test_mistralai_chat_completion.py	Adds finish-reason mapping regression tests.
python/​semantic_kernel/​connectors/​ai/​mistral_ai/​services/​mistral_ai_chat_completion.py	Centralizes safe finish-reason conversion for both response paths.

_{Review effort: Balanced}