← All benchmarks

APEX-Agents (release-table score, metric unspecified)

Professional Work

APEX-Agents values from release tables that do not state the metric, version or harness, and whose comparator values do not match Mercor's 480-task pass@1 leaderboard (for example the Grok 4.6 launch table). Diagnostic until the metric is resolved; never substituted for APEX-Agents pass@1.

Top models (higher is better)