Changelog

Opik Dashboard:

• Improved the UX when navigating between the project list page and the traces page

Python SDK:

• Make the logging of spans and traces optional when using Opik LLM metrics

JS / TS SDK:

• Added logs and better error handling

Opik Dashboard:

• Added support for local models in the Opik playground

Python SDK:

• Improved the @track decorator to better support nested generators.
• Added a new Opik.copy_traces(project_name, destination_project_name) method to copy traces from one project to another.
• Added support for searching for traces that have feedback scores with spaces in their name.
• Improved the LangChain and LangGraph integrations

JS / TS SDK:

• Released the Vercel AI integration
• Added support for logging feedback scores

Opik Dashboard:

• You can now view feedback scores for your projects in the Opik home page
• Added line highlights in the quickstart page
• Allow users to download experiments as CSV and JSON files for further analysis

Python SDK:

• Update the evaluate_* methods so feedback scores are logged after they computed rather than at the end of an experiment as previously
• Released a new usefulness metric
• Do not display warning messages about missing API key when Opik logging is disabled
• Add method to list datasets in a workspace
• Add method to list experiments linked to a dataset

JS / TS SDK:

• Official release of the first version of the SDK - Learn more here
• Support logging traces using the low-level Opik client and an experimental decorator.

Opik Dashboard:

• Performance improvements for workspaces with 100th of millions of traces
• Added support for cost tracking when using Gemini models
• Allow users to diff prompt

SDK:

• Fixed the evaluate and evaluate_* functions to better support event loops, particularly useful when using Ragas metrics
• Added support for Bedrock invoke_agent API

Opik Dashboard:

• Added logs for online evaluation rules so that you can more easily ensure your online evaluation metrics are working as expected
• Added auto-complete support in the variable mapping section of the online evaluation rules modal
• Added support for Anthropic models in the playground
• Experiments are now created when using datasets in the playground
• Improved the Opik home page
• Updated the code snippets in the quickstart to make them easier to understand

SDK:

• Improved support for litellm completion kwargs
• LiteLLM required version is now relaxed to avoid conflicts with other Python packages

Opik Dashboard:

• Datasets are now supported in the playground allowing you to quickly evaluate prompts on multiple samples
• Updated the models supported in the playground
• Updated the quickstart guides to include all the supported integrations
• Fix issue that means traces with text inputs can’t be added to datasets
• Add the ability to edit dataset descriptions in the UI
• Released online evaluation rules - You can now define LLM as a Judge metrics that will automatically score all, or a subset, of your production traces.

SDK:

• New integration with CrewAI
• Released a new evaluate_prompt method that simplifies the evaluation of simple prompts templates
• Added Sentry to the Python SDK so we can more easily

Opik Dashboard:

• Fixed an issue with the trace viewer in Safari

SDK:

• Added a new py.typed file to the SDK to make it compatible with mypy

Opik Dashboard:

• Added duration chart to the project dashboard
• Prompt metadata can now be set and viewed in the UI, this can be used to store any additional information about the prompt
• Playground prompts and settings are now cached when you navigate away from the page

SDK:

• Introduced a new OPIK_TRACK_DISABLE environment variable to disable the tracking of traces and spans
• We now log usage information for traces logged using the LlamaIndex integration

SDK:

• Improved error messages when getting a rate limit when using the evaluate method
• Added support for a new metadata field in the Prompt object, this field is used to store any additional information about the prompt.
• Updated the library used to create uuidv7 IDs
• New Guardrails integration
• New DSPY integration

Opik Dashboard:

• The Opik playground is now in public preview

  

• You can now view the prompt diff when updating a prompt from the UI
• Errors in traces and spans are now displayed in the UI
• Display agent graphs in the traces sidebar
• Released a new plugin for the Kong AI Gateway

SDK:

• Added support for serializing Pydantic models passed to decorated functions
• Implemented get_experiment_by_id and get_experiment_by_name methods
• Scoring metrics are now logged to the traces when using the evaluate method
• New integration with aisuite
• New integration with Haystack