AI pricing and market data you can verify.

Choosing a model means checking a vendor’s pricing page, a marketplace listing, a model card, a leaderboard and a policy document — then repeating it when any of them changes. The figures rarely agree, and nothing tells you which one is current. StackTicker collects them in one place and keeps the disagreements visible rather than picking a winner quietly.

Prices are read directly from provider APIs and vendor pricing pages. Specifications come from vendor documentation. Benchmark results come from the organisations that published them, and are attributed to those organisations rather than to us — we do not run these evaluations.

Tracked market data — every price, rate and benchmark result — opens to show the URL it came from, the date it was read and its verification state. Figures we derive from those, such as counts and price spreads, name the values they were computed from instead. Where two sources disagree, both are kept, and the model’s own publisher takes precedence over a marketplace reselling it.

StackTicker does not invent a missing source value or replace it with zero. A model missing a benchmark was not measured, and reporting that as a low score would be a claim nobody made.

When a value is calculated by StackTicker — such as a normalized rate, ranking or workload estimate — it is labeled as a calculation and remains tied to the source data used to produce it.

Coverage varies by model and provider, and each page shows what information is available.

StackTicker helps developers, teams and businesses compare AI models, providers and compute using current published data and traceable sources.