Multi-select filters, track status codes, benchmark latency and token usage + more

Another week at Athina: a new bundle of improvements to the platform.

Observe

1. Multi-select filters in the Dashboard, Analytics, and Compare pages

2. Support for tracking status codes and errors in LLM requests

3. Include chat history in custom prompt evals

4. Ability to split custom attributes into separate columns when creating a dataset

+++ many minor improvements

Develop

4. Support for structured outputs, JSON mode and tool calling in Dynamic Columns, Experiments and Prompt Playground.

5. View response times and token usage when running a Dynamic Column or Experiment

6. OpenAI Assistant is now supported as a new dynamic column

7. You can now rename prompts when configuring an experiment

+++ many minor improvements

+++

🐛 Note: Countless bugs were killed in the making of this week at Athina.