The full table

Every model, in the open

Sortable by any column, filterable by use case — sourced or blank, never guessed.

All models

Every model, sortable by any column and filterable by use case. Coding score is 0–100 — SWE-bench Verified where published, otherwise a sourced estimate (marked est); the dot + label show confidence. Click a row for the basis, verdict, and sources. "—" means the figure wasn't publicly sourced. We don't guess.

Model Best for Codingscore /100 Context Pricein / out · 1M