SaaSBench (Engineering) - Deploy Capability: leaderboard
The Deploy capability dimension of SaaSBench (Ren et al., 2026), pooled over all 30 SaaS systems a coding agent builds from scratch: whether the built service starts cleanly, with dependencies, schema migrations, configuration, seeding and health checks. Share of the validation-node score earned in that dimension, in percent; higher is better.
Metric: Capability score (%; share of the Deploy (service start, migrations, configuration, health checks) validation-node max score earned across all 30 tasks, pass@1; agent builds and deploys an enterprise SaaS system from a PRD). Source: arxiv.org. Saturation forecast: Around 2030. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 22.76 | #75 |
| 2 | DeepSeek V4 Pro | 21.31 | #181 |
| 3 | GPT-5.4 (xHigh) | 19.75 | #121 (GPT-5.4) |
| 4 | GLM-5.1 | 19.25 | #175 |
| 5 | Kimi K2.6 | 18.68 | #173 |
| 6 | MiniMax-M2.7 | 18.03 | #519 |
| 7 | Gemini 3.1 Pro (Preview) | 17.69 | #91 |
| 8 | Qwen 3.6 Plus | 14.12 | #217 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=saasbench-engineering-deploy-capability · How It Works · Data refreshed daily, snapshot 2026-11-12.