{% extends "base.html" %} {% block title %}Harness x Model Matrix — AgentArena{% endblock %} {% block content %}

Harness x Model Matrix

Same LLM, different harnesses — who works cleaner, faster, cheaper?

{% if harnesses and models %}
{% for model in models %} {% endfor %} {% for harness in harnesses %} {% for model in models %} {% set key = harness ~ "|" ~ model %} {% set cell = cells.get(key) %} {% endfor %} {% endfor %}
Harness{{ model }}
{{ harness }} {% if cell %}
{{ "%.0f"|format(cell.pass_at_1 * 100) }}%
pass@1
{% if cell.avg_composite is not none %}
{{ "%.0f"|format(cell.avg_composite * 100) }}% comp
{% endif %}
${{ "%.4f"|format(cell.avg_cost) }} {{ "%.0f"|format(cell.avg_duration) }}s {{ cell.total_runs }}r
{% else %} — {% endif %}
Legend
70%+ = strong 40-69% = moderate <40% = weak pass@1 = reliability comp = composite score r = total runs
{% else %}

No data yet

Run benchmarks with different harness/model combinations to populate the matrix.

Go to Leaderboard
{% endif %} {% if harnesses %}

Harness Reports

{% for harness in harnesses %} {{ harness }} {% endfor %}
{% endif %} {% endblock %}