Coding Agent Intelligence, Cost, and Token Explorer

christiant.io

Christian Taillon

Coding Agent Intelligence, Cost, and Token Explorer

Compare selected frontier coding-agent configurations across intelligence, cost, and token use.

Configurations shown
0
Requested model families
Greatest Pareto outlier
--
--
Lowest cost per task
--
--
Median token use
--
Average total tokens per task
20%67%
$0.05$12

3D scatter cube

Drag to rotate, scroll to zoom, and select a point for details.

No configurations match the current filters.

Graph costs reflect pre-reduction pricing. OpenAI cut GPT-5.6 Luna by 80% and Terra by 20% on July 30, 2026 (Sol unchanged) — details. The points, Pareto frontier, and rankings here have not been updated.

Select a configuration

IntelligenceHover or click a point
Cost per task--
Tokens per task--

Model company legend

Intelligence uses the Artificial Analysis Coding Agent Index. Cost and token utilization are per-task averages from the uploaded coding-agent dataset. Colors identify the model company; harness is shown in hover and selection details only. Current-generation models and all model companies are enabled by default; use “Include older models” to reveal earlier Opus, GPT, Kimi, GLM, Composer, and related generations. GPT-5.6 per-task costs reflect pre-reduction pricing; OpenAI lowered Luna by 80% and Terra by 20% on July 30, 2026 (Sol unchanged).