InferenceBench (Prompt v1.0.1): leaderboard
Metric: Geometric Mean Speedup. Source: inferencebench.ai. Saturation forecast: Around March 2028. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 5.5 (Max) | 12.08 |
| 2 | Claude Fable 5.1 | 9.83 |
| 3 | Claude Fable 5.1 (Max) | 9.31 |
| 4 | Claude Opus 5 | 8.9 |
| 5 | Claude Fable 5 (Low) | 8.74 |
| 6 | Claude Fable 5 | 7.52 |
| 7 | GLM-5.2 (Max) | 7 |
| 8 | Claude Sonnet 5 | 6.43 |
| 9 | Kimi K2.7 Code | 6.27 |
| 10 | Kimi K3 | 5.7 |
| 11 | GLM-5.3 (Max) | 4.98 |
| 12 | Muse Spark 1.3 | 4.53 |
| 13 | GLM-5.3 Flash | 4.49 |
| 14 | Grok 4.5 | 3.42 |
Interactive version: theaggregate.ai/benchmark?slug=inferencebench-prompt-v1-0-1 · How It Works · Data refreshed daily, snapshot 2026-10-09.