ModelCards
← All benchmarks

BFCL v3

We haven't written a description for this benchmark yet; see the card notes below for how each maker ran it.

Your workloadRAG over docs · 12K in · 600 out · 50K requests/mo · 30% cached
  1. 1MAI-Thinking-1 · Microsoftbest value
    72.0%

“Best value”: no other model here reports a higher score at a lower cost for your workload. Scores are self-reported by each maker with different settings, so treat close calls as ties.

How each maker reported it

  • MAI-Thinking-1 Tables 11–12, p.53–54

    Microsoft's own runs, average of 4 at T=1, top-p 0.97. Other models' scores in the first table come from their official cards. Sonnet 4.6 values for MMLU-Pro and later rows were run by Microsoft at max effort. Blank = not reported.

Only each model's own maker's number is shown; scores a maker reports for rival models are left out.