Benchmark Almanac

Arize AI

We track 3 benchmarks that Arize AI made or helped make.

0 in use 3 with public code 2 with a leaderboard here

Benchmarks

Name Kind Released Best score Status
Cost per Successful Task Agents solve 40 Terminal-Bench tasks. The benchmark divides spending across all attempts by the number of successful runs. Terminal tasks Jul 2026 $23 Not in cards
Acme Agent Evals Agents use GitHub tools to answer questions about issues, pull requests, and code in a simulated software project. Apps and MCP servers May 2026 85.1% Not in cards
Decision Model Benchmark Models classify examples of hallucinations, judge model outputs, and route agent requests. Hallucination rates Oct 2026 87.9% Not in cards