SQBench - L1 Atomic Capabilities: leaderboard
Metric: Strict Pass@1 (%; the 100 L1 atomic-capability tasks on SQBench v1.0, 220 production-oriented delivery tasks in which an agent processes input assets, uses tools and produces a specified deliverable; a task passes only with functional Completion 1 and no triggered 10-dimension risk penalty; one run per configuration and task; higher is better). Source: arxiv.org. Saturation forecast: Around July 2027. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.8 (Max) | 63 |
| 2 | Kimi K3 | 61 |
| 3 | Qwen 3.7 Max | 52 |
| 4 | GLM-5.2 (Max) | 50 |
| 5 | Gemini 3.5 Flash (High) | 49 |
| 6 | DeepSeek V4 Pro (Max) | 46 |
| 7 | Qwen 3.7 Plus | 45 |
| 8 | Hy3 | 43 |
| 9 | Qwen 3.6 Plus | 42 |
| 10 | GLM-5.1 | 42 |
| 11 | Kimi K2.7 Code | 42 |
| 12 | MiMo-V2.5 | 41 |
| 13 | Kimi K2.6 | 40 |
| 14 | DeepSeek V4 Pro (High) | 40 |
| 15 | Doubao-Seed-2.1-Pro (High) | 39 |
Interactive version: theaggregate.ai/benchmark?slug=sqbench-l1-atomic-capabilities · How It Works · Data refreshed daily, snapshot 2026-09-29.