Aeon AI Risk Management

Model Economics Lab

Compare coding intelligence, task cost, and token consumption across model and reasoning-effort settings. Filter the master plot, inspect point-level evidence, and download the dataset.

Executive finding

Buy reasoning effort by task class, not by brand. Luna high is the strongest routine-value point, Terra max compresses frontier cost, and Sol max should be reserved for tasks that can repay its cost jump.

Master plot

The horizontal axis is cost per coding task, the vertical axis is Artificial Analysis Coding Agent Index, and bubble area represents output tokens per task. Filled points are measured, while hollow points are explicitly modeled.

Interactive filters

Users can filter models, reasoning-effort levels, measured or modeled evidence, toggle trajectories, and switch the cost axis between logarithmic and linear scales.

Measured anchors

Sol max reaches 80.0 at $7.08 per task. Terra max reaches 77.4 at $2.76. Luna high reaches 67.9 below $1. Grok 4.5 high reaches 76.4 at $2.59. Fable max reaches 77.2 at $11.75.

Effort economics

Higher effort creates nonlinear token and cost growth. The last capability points can be valuable for ambiguous, long-horizon, or high-consequence work, but are usually inefficient for routine coding.

Modeled points

Where no comparable public run exists, Aeon uses neighboring model-family effort curves or a vendor-published effort curve. Every estimate is marked as modeled and is not presented as a benchmark result.

Limitations

Output tokens do not include every input, cache, retry, tool, and harness cost. Coding-agent token use can vary substantially by task and harness, so organizations should validate routing against their own repositories and acceptance tests.

Downloadable dataset

The full operating-point dataset is available in CSV and JSON with model, provider, effort, index, cost, output tokens, evidence status, modeling basis, and source URL.

Update policy

Aeon appends new models and effort settings when comparable public runs and point-level provenance become available, then publishes a concise research update.

Operating recommendation

Use a routing ladder that assigns cheaper effort to bounded tasks and reserves frontier settings for ambiguity, long dependency chains, and changes with high failure cost.