AI Models Comparison

Comparison Matrix Updated October 21, 2026
Baseline: Claude Code + Sonnet 5.5 Medium = 100% Quality / 100% Limit / 100% Reliability
Green quality = superior to baseline Green limit = more economical than baseline Green reliability = higher autonomy

Haiku 5.5: preliminary editorial estimates with low confidence. Quality/reliability use a proxy model, not measured ratios. A dash means no data; subscription resource consumption is unknown.

Gemini 4 Argon High: estimated resource index 235–285, anchored to Gemini 3.1 Pro. Low confidence; assumes introductory rates apply. Without the discount: about 470–565. These are reconstructed estimates, not measured subscription debits.

Each available metric of the selected configuration equals 100. Each category is recalculated from the original estimates: this changes the comparison scale, not the model ratings. Your choice is saved in this browser.

Normalized estimates, not measured benchmarks. Resource index: lower is better; it is not an official subscription quota.

Environment:
Environment Model Effort Quality Quality ↕ Limit usage Limit ↕ Reliability Reliability ↕