GBA Eval - Audio: leaderboard
GBA Eval section score for emulated PSG and FIFO audio, graded by log-mel spectrogram distance from the Mesen2 reference on a 0-1 scale, higher is better. Carries 20% of the overall GBA Eval score; ungraded emulators are omitted.
Metric: PSG and FIFO audio score from log-mel spectrogram distance (0-1). Source: gbaeval.com. Saturation forecast: Around February 2027. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 0.69 | #39 |
| 2 | Claude Opus 5 | 0.66 | #16 |
| 3 | Claude Sonnet 5 | 0.64 | #65 |
| 4 | GPT-5.5 | 0.59 | #33 |
| 5 | Grok 4.5 | 0.48 | #54 |
| 6 | Claude Fable 5 | 0.45 | #20 |
| 7 | GPT-5.6 Sol | 0.4 | #22 |
| 8 | Kimi K3 | 0.37 | #35 |
| 9 | GPT-5.4 | 0.37 | #83 |
| 10 | Claude Sonnet 4.6 | 0.17 | #94 |
| 11 | Claude Opus 4.7 | 0.1 | #51 |
| 12 | Grok Build 0.1 | 0.07 | #137 |
| 13 | Claude Opus 4.6 | 0.01 | #67 |
| 14 | Gemini 3.1 Pro (Preview) | 0 | #64 |
| 15 | Kimi K2.6 | 0 | #113 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gba-eval-audio · How It Works · Data refreshed daily, snapshot 2026-11-05.