About modhub.ai

The AI model market moves faster than any one team can track. New flagships ship monthly, prices get cut overnight, and benchmark results scatter across model cards, blog posts and papers. modhub.ai exists to compress that chaos into one honest, comparable, up-to-date picture — free, with no account required.

We track 51 models from 13 providers across 7 benchmarks, alongside API pricing, context windows, licenses and modalities. Everything on this site is also available through the free JSON API.

Methodology

Benchmark scores. Figures come from official model cards and provider announcements wherever available. When a number is community-reported, derived from third-party evaluations, or normalized across benchmark versions, it is flagged as an estimate and rendered with a ~ prefix — in tables and in the API (isEstimate: true). We never silently mix estimated and official figures.

The modhub Index.A composite score equal to the plain average of a model's results across all tracked benchmarks (all are 0–100 scales). A model needs results on at least 4benchmarks to qualify, so sparse coverage can't inflate a ranking. No weighting, no opinion — it is fully reproducible from the public API in one line of code.

Pricing.We list standard per-token API prices from the model's first-party provider; for open-weights models without a first-party API, a representative major host. Long-context surcharges, batch discounts and enterprise agreements are noted but not folded into headline prices. Cached input pricing is tracked separately because it dominates real agent economics.

Data & updates. The dataset is reviewed by hand — last update 2026-06-10. Models without published pricing or scores show explicit gaps rather than guesses. Spotted an error? We treat data corrections as the highest-priority issue class.

Independence

modhub.ai is independent and not affiliated with any AI lab. We don't accept payment for rankings; if sponsored placements ever appear, they will be unambiguously labeled and will never affect scores or ordering.