GBA Eval
Frontier coding agents get 24 hours to write a complete Game Boy Advance emulator (Rust + WebAssembly) from scratch, graded against the Mesen2 reference emulator.
What this benchmark measures
Frontier coding agents get 24 hours to write a complete Game Boy Advance emulator (Rust + WebAssembly) from scratch, graded against the Mesen2 reference emulator.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is overall score. It should be interpreted within GBA Eval, not compared as part of a site-wide ranking.
What to be careful about
Overall = 0.60 replay (SSIM) + 0.20 audio (log-mel) + 0.20 procedural; grading date differs per model run, so keep it attached.
Frequently asked
What is GBA Eval?
Frontier coding agents get 24 hours to write a complete Game Boy Advance emulator (Rust + WebAssembly) from scratch, graded against the Mesen2 reference emulator. It is a coding benchmark measured by overall score.
What does overall score mean on GBA Eval?
GBA Eval reports overall score (%); higher is better. Scores are shown only within GBA Eval and are never averaged with other benchmarks.
What is the top reported GBA Eval score?
Claude Opus 4.8 has the top reported score on GBA Eval: 70.9% (overall score).
Why do GBA Eval scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. GBA Eval scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".