Benchmarks / Coding
GBA Eval
Uses the first-party leaderboard JSON overall emulator score. Score: overall emulator score. Collected by Jake Boggs' benchmark dashboard.
GBA Eval
Higher is better · best effort per model
full table below
| # | Model | Effort | Score |
|---|---|---|---|
| 1 | Opus 5 | max | 79.6% |
| 2 | Fable 5 | max | 74.5% |
| 3 | Opus 4.8 | max | 70.9% |
| 4 | Grok 4.5 | max | 65.4% |
| 5 | Sonnet 5 | max | 65.3% |
| 6 | GPT-5.5 | xhigh | 53.2% |
| 7 | GPT-5.6 Sol | max | 52.6% |
| 8 | Sonnet 4.6 | max | 48.8% |
| 9 | Kimi K3 | max | 48.3% |
| 10 | GPT-5.4 | xhigh | 45.1% |
| 11 | Opus 4.6 | max | 44.1% |
| 12 | Opus 4.7 | max | 43.8% |
| 13 | Muse Spark 1.1 | max | 7.9% |
| 14 | Gemini 3.5 Flash | xhigh | 6.7% |
| 15 | Grok Build 0.1 | – | 2.4% |
| 16 | MiniMax M3 | max | 0.9% |
| 17 | Kimi K2.6 | max | 0.9% |
| 18 | Kimi K2.7 | none | 0.9% |
| 19 | Gemini 3.1 Pro | high | 0.8% |
| 20 | Qwen 3.7 Max | max | 0.4% |
| 21 | GLM 5.2 | max | 0.0% |
| 22 | GLM 5.1 | max | 0.0% |
| 23 | MiniMax M2.7 | high | 0.0% |