Skip to main content

​
Default Playground Provider and Model

Available in arize-phoenix 15.6.0+ You can now save a personal default provider and model for the Playground from the AI Providers settings page. Phoenix stores the preference in your browser and uses it whenever you open a new Playground session. When no preference is set, Playground falls back to the existing default (OpenAI / gpt-4o). If you’ve previously configured per-provider invocation parameters (model, temperature, max tokens), those saved settings still take precedence for your preferred provider — the preference acts as a starting point, not an override.

​
Span Metrics Always Visible

Available in arize-phoenix 15.6.0+ The metrics aside — showing latency percentiles, token counts, and error rates — is now always visible on the right side of the spans table on the project page.