SaaS-Bench: leaderboard
Metric: Overall (self-reported). Source: benchmarklist.com. 17 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 43.9 |
| 2 | GPT-5.5 (High) | 43.8 |
| 3 | Claude Opus 4.6 | 43.2 |
| 4 | GPT-5.4 (High) | 37 |
| 5 | Kimi K2.6 | 34.1 |
| 6 | Qwen 3.6 Plus | 29.9 |
| 7 | Kimi K2.5 | 27.7 |
| 8 | Gemini 3.1 Pro (Preview) | 27.1 |
| 9 | Seed 2.0 Pro | 27.1 |
| 10 | Step 3.7 Flash | 25.1 |
| 11 | Claude Sonnet 4.6 | 23.3 |
| 12 | Gemini 3.5 Flash (High) | 23.3 |
| 13 | DeepSeek V4 Pro | 21.5 |
| 14 | GLM-5.1 | 17.4 |
| 15 | MiniMax-M2.7 | 15.8 |
Interactive version: theaggregate.ai/benchmark?slug=saas-bench · How It Works · Data refreshed daily, snapshot 2026-09-05.