BizAgentBench - Strict Success: leaderboard
BizAgentBench / BizBench (2026), 50 long-horizon banking tasks with deterministic final-state and trajectory requirements, revised customer simulator protocol. Claude Opus 5 and Muse Spark 1.3 have one run per task; other models have three, averaged. Goal attainment, process compliance, protected constraints and joint strict success are distinct outcomes; intermediate failure is failure of process or protection. Rate (%); higher is better.
Metric: Rate (%). Source: arxiv.org. Saturation forecast: Around January 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 | 64 | #12 |
| 2 | GPT-6 Sol | 61.3 | #14 |
| 3 | GLM-5.3 Flash | 43.3 | #65 |
| 4 | DeepSeek V4.1 Flash | 43.3 | #36 |
| 5 | DeepSeek V4 Pro | 39.3 | #102 |
| 6 | Muse Spark 1.3 | 38 | #24 |
| 7 | Qwen 3.8 27B | 32.7 | #105 |
| 8 | DeepSeek V4 Flash | 5.3 | #137 |
| 9 | Qwen 3.5 397B A17B | 0.7 | #146 |
| 10 | Qwen 3.5 9B | 0 | #379 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=bizagentbench-strict-success · How It Works · Data refreshed daily, snapshot 2026-10-20.