Accord PartnersAI Benchmark

Which model for which task

An evidence-based view of frontier and open-source AI models — refreshed monthly, mapped to the work that gets done at an audit firm. Recommendations balance performance and cost, with performance taking priority.

Decision ledger — monthly findings

    Pricing reference

    Token rates verified this audit. Open-source rates reflect API-as-a-service from primary providers; self-hosting is 5–20× cheaper but adds operations burden.

    VendorModelLicense$ / 1M input$ / 1M output

    Methodology

    How this report is built and what to trust.

    Principles

      Sources

        Refresh cadence