APIFlow-Bench 1.0 - 20-Subtask Chain: leaderboard
Postman's APIFlow-Bench 1.0 measures whether agents complete long, dependent REST API workflows (arXiv 2608.29128v1). This board covers the leaderboard's default 20-subtask chain set (11 tasks); pass rate is passed over all trials, 0-100, higher is better, and refusals and abstentions count as failures. From the live leaderboard's overview page.
Metric: Pass Rate (%). Source: www.postman.com. Saturation forecast: Around 2029. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 72.73 | #33 |
| 2 | Kimi K3 | 72.73 | #35 |
| 3 | Claude Opus 4.8 | 69.09 | #39 |
| 4 | DeepSeek V4 Flash | 69.09 | #155 |
| 5 | GLM-5.2 | 65.45 | #80 |
| 6 | Qwen 3.7 Plus | 65.45 | #98 |
| 7 | DeepSeek V4 Pro | 63.64 | #112 |
| 8 | GPT-5.6 Sol | 63.64 | #22 |
| 9 | Qwen 3.6 Plus | 63.64 | #124 |
| 10 | Claude Haiku 4.5 (20251001) | 63.64 | #264 |
| 11 | GPT-5.6 Luna | 63.64 | #79 |
| 12 | GPT-5.6 Terra | 63.64 | #40 |
| 13 | Claude Sonnet 4.6 | 61.82 | #94 |
| 14 | Kimi K2.7 Code | 61.82 | #95 |
| 15 | Claude Sonnet 5 | 60 | #65 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=apiflow-bench-1-0-20-subtask-chain · How It Works · Data refreshed daily, snapshot 2026-11-05.