Benchmarks

Independent standards for the AI work that touches employees.

Each PYX Labs benchmark defines what "good" looks like in a real workplace domain — scored against criteria written by the I-O psychologists who do the work, and graded by a cross-vendor judge.

FlagshipLive

PYX-Voice

How well frontier models do employee listening — analyzing and interpreting employee data to move the needle on organizational performance.

Task types
2 — Verifiable response + Summary
Evaluation criteria
208 discrete criteria
Capabilities tested
7 LLM capabilities
Analysis types
7 types included
Experience factors
10 employee-experience factors
Difficulty levels
3 levels
Models tested
7 frontier models
Explore benchmark →MethodologyRead the paper
Expert-graded scale
Validated results
76.0
71.0
70.0
68.0
66.0
65.0
54.0
Expert-graded 0–100 scale. Model identities anonymized.
Live webinar

See PYX-Voice in action — a live walkthrough of the benchmark, the tasks, and the results.

Save your spot→
On the roadmap

More standards, in development.

Roadmap is directional and subject to change.
EX-02Coming

PYX-Coach

AI coaching

Whether an AI coach gives guidance that is grounded, responsible, and genuinely useful — judged against how expert coaches handle the same conversations.

EX-03Coming

PYX-Learn

AI learning tools

Whether AI-based learning tools teach accurately and adapt to the learner — measured against instructional-design and learning-science standards.