Skip to main content
XOR

gemini-3-1-pro-preview

run under the gemini31 harness

The harness is part of what is measured here: the same model under a different runner scores differently. Why that matters.

58.7%

repair rate

44–71

95% interval

4

rank of 15

$3.34

per repair · estimated

Rank is a position in a table whose intervals overlap almost everywhere. Read it as a band, not a place.

outcome mix

64 repaired · 18 not repaired · 27 did not build

64 repaired · 18 not repaired · 27 did not build

By weakness class

WeaknessRepair rateRateAttempts
CWE-125Out-of-bounds Read66.2% (95% interval 50.0–79.0)66.2%65
CWE-787Out-of-bounds Write62.5% (95% interval 42.1–81.3)62.5%16
CWE-416Use After Free33.3% (95% interval 5.9–70.0)33.3%12
CWE-908Use of Uninitialized Resource0.0% — no interval: too few codebases or attempts to generalise0.0%4
CWE-843Access of Resource Using Incompatible Type ('Type Confusion')50.0% — no interval: too few codebases or attempts to generalise50.0%2
CWE-476NULL Pointer Dereference50.0% — no interval: too few codebases or attempts to generalise50.0%2
CWE-415Double Free100.0% — no interval: too few codebases or attempts to generalise100.0%2
CWE-1284Improper Validation of Specified Quantity in Input100.0% — no interval: too few codebases or attempts to generalise100.0%1
CWE-475100.0% — no interval: too few codebases or attempts to generalise100.0%1
CWE-5620.0% — no interval: too few codebases or attempts to generalise0.0%1

Compare

vs codex/gpt-5-2vs cursor/opus-4-6vs claude/claude-opus-4-6vs opencode/gemini-3-1-pro-preview

Corpus cve-bench-136 · run February 2026 · How we test