Phoenix v20.21.0 adds Mistral AI cost tracking
Updated Phoenix
Summary
Arize Phoenix v20.21.0, released October 11, 2026, adds Mistral AI model pricing to cost tracking and fixes evaluator metadata loss.
- Tool
- Phoenix
- Category
- Agent infra
- Date
- Models
- Mistral AI
- Pricing
- Not stated in the sources we cited.
Arize published Phoenix v20.21.0 on October 11, 2026, according to the GitHub release. The release adds Mistral AI model pricing to cost tracking and includes a fix for evaluator metadata that was being dropped.
Phoenix is described in its repository as "AI Observability & Evaluation", and sits in the Agent infra category here.
Release facts
| Fact | Detail |
|---|---|
| Product | Phoenix (Arize-ai/phoenix) |
| Release | arize-phoenix v20.21.0 (tag arize-phoenix-v20.21.0) |
| Release date | October 11, 2026 |
| License | Not stated in the cited sources; the GitHub API reports no recognized license |
| Where to get it | GitHub release page and the Phoenix docs at arize.com/docs/phoenix |
What changed
- Mistral AI pricing: cost tracking now supports Mistral AI model pricing, per the release notes.
- Evaluator metadata: the release stops dropping evaluator metadata nested under
result. - Evals dependency: arize-phoenix-evals is updated to 3.9.2.
- Playground: built-in model prices and model lists are synced.
- Agents security: pydantic-ai is upgraded for CVE-2026-107286.
- Projects: projects created by ingestion get a random gradient.
Questions
Does this release add support for Mistral models?
The release notes list "support Mistral AI model pricing" under cost-tracking features. They do not describe other Mistral-specific changes.
Does it include a security fix?
According to the release notes, the agents component upgrades pydantic-ai for CVE-2026-107286. The notes do not give further detail.
Is the license stated?
No. The repository page says the GitHub API does not report a recognized license, so check the repository before relying on any terms.
See the Phoenix hub, the Agent infra category, or the related Opik 2.2.97 launch for comparable observability releases.
Alternatives
| Tool | License | Pricing note |
|---|---|---|
PhoenixThis launchPhoenix is an Arize-maintained AI observability and evaluation tool with prompt and dataset tooling. | — | Not stated |
LangfuseOpen source platform for tracing, evaluating and improving AI agents and LLM applications. | MIT | Not stated |
OpikOpen-source platform for tracing, evaluating and optimizing LLM applications and agents. | Apache-2.0 | Not stated |
Questions
What is Phoenix?
The GitHub repository Arize-ai/phoenix is described as "AI Observability & Evaluation". Its docs list features such as LLM-based evaluations, prompt management, a prompt playground, span replay, datasets and experiments.
Is Phoenix open source?
Not stated in the cited sources. The excerpts say the GitHub API does not report a recognized license for the repository, so no license is claimed here.
How is Phoenix used?
According to its docs, you can create datasets by collecting traces or uploading from code/CSV, run experiments to compare different versions of your application, and attach evaluators to datasets so they run automatically during experiments.
Filed on Agent Radar. Claims in this note are limited to the sources above.