Benchmarks · Agents & computer use
StoreBench
Uses the mean composite score on the 11-task main suite, excluding human and scripted baselines and the separate full-year experiment. Score: mean composite score. Collected by Jake Boggs' benchmark dashboard.
13 models · 13 results · data as of 2026-11-15 · afterquery.com ↗
Leaderboard
Fable 5.2
All results
13 rows| # | Model | Effort | StoreBench |
|---|---|---|---|
| 1 | Fable 5.2 | 83.5% | |
| 2 | GPT-6 Astra | 77.2% | |
| 3 | Opus 5.5 | 76.2% | |
| 4 | Fable 5.1 | 72.5% | |
| 5 | GPT-6.1 Sol | 72.5% | |
| 6 | DeepSeek V4 Pro | 70.0% | |
| 7 | Gemini 4 Argon | 67.7% | |
| 8 | Opus 4.8 | 66.8% | |
| 9 | Fable 5 | 62.9% | |
| 10 | Qwen 3.8 Max | 58.7% | |
| 11 | Gemini 3.8 Flash | 55.2% | |
| 12 | GPT-5.6 Sol | 53.8% | |
| 13 | Muse Spark 1.2 | 38.7% |