fix(cohere): use Cohere V2 API to support latest models

[phlogisticfugu]

• support V1 and V2 Cohere SDK clients (default to V2)
• pass model explicitly in calls (fixing existing issue)
• update tests and examples to use V2
• fix ollama cohere models which support tool calling
• add comments for more models

---

Important

Update Cohere integration to support V1 and V2 API clients, defaulting to V2, with updated tests and examples.

• Behavior:
  • Support for both V1 and V2 Cohere SDK clients in instructor/providers/cohere/client.py, defaulting to V2.
  • Explicit model specification in API calls to fix existing issues.
  • Updated response handling to accommodate V2 API structure in instructor/processing/function_calls.py and instructor/providers/cohere/utils.py.
• Testing:
  • Updated tests in tests/llm/test_cohere/test_json_schema.py and tests/llm/test_cohere/test_none_response.py to use V2 API.
  • Added fixtures for V1 and V2 clients in tests/llm/test_cohere/conftest.py.
• Documentation:
  • Updated examples in docs/examples/document_segmentation.md and docs/integrations/cohere.md to use V2 API.
• Misc:
  • Minor code formatting changes in examples/caching/run.py and instructor/processing/multimodal.py.

^{This description was created by for cee3a6d. You can customize this summary. It will automatically update as commits are pushed.}

[ellipsis-dev]

Important

Looks good to me! 👍

Reviewed everything up to cee3a6d in 1 minute and 32 seconds. Click for details.

• Reviewed 965 lines of code in 23 files
• Skipped 0 files when reviewing.
• Skipped posting 6 draft comments. View those below.
• Modify your settings and rules to customize what types of comments Ellipsis leaves. And don't forget to react with 👍 or 👎 to teach Ellipsis.

1. tests/test_process_response.py:7

• Draft comment:
  In test_typed_dict_conversion(), adding a comment explaining why the TypedDict and Pydantic model conversions are equivalent would clarify the assumptions for future maintainers.
• Reason this comment was not posted:
  Confidence changes required: 70% <= threshold 85% None

2. tests/test_process_response.py:22

• Draft comment:
  In test_openai_to_bedrock_conversion(), an inline comment explaining the transformation logic (e.g. removal of 'model' and creation of 'modelId', as well as the expected two messages) would improve clarity.
• Reason this comment was not posted:
  Confidence changes required: 60% <= threshold 85% None

3. tests/test_process_response.py:45

• Draft comment:
  In test_bedrock_round_trip(), add a comment noting that preserving the original input via deep copy is critical for ensuring round-trip integrity.
• Reason this comment was not posted:
  Confidence changes required: 60% <= threshold 85% None

4. tests/test_process_response.py:109

• Draft comment:
  In test_empty_and_missing_content(), consider documenting any expected side effects (like logging warnings) when handling empty messages or missing content.
• Reason this comment was not posted:
  Confidence changes required: 50% <= threshold 85% None

5. tests/test_process_response.py:123

• Draft comment:
  In test_bedrock_invalid_content_format(), using pytest.raises with the match parameter to assert the specific error message would be clearer than manually raising an AssertionError.
• Reason this comment was not posted:
  Confidence changes required: 70% <= threshold 85% None

6. instructor/providers/cohere/utils.py:209

• Draft comment:
  Typo: The comment references 'handle_cohere_modes', but the function is named 'handle_cohere_tools'. Please verify and correct if needed.
• Reason this comment was not posted:
  Comment was not on a location in the diff, so it can't be submitted as a review comment.

Workflow ID: wflow_9jLdYa2oMPpJ9r4D

^{You can customize by changing your verbosity settings, reacting with 👍 or 👎, replying to comments, or adding code review rules.}