AIM Routing
Agents
Zeta's private tool-routing benchmark for its marketing agents: 475 marketing-analytics requests, each paired with the set of tools or reports needed to answer it. A request passes only when the model's complete set of selected tool names exactly equals the expected set, so a missing or extra tool fails it; the score is the share of requests passed. Zeta runs it and supplied the results to Fireworks' Specialized Intelligence Index.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6.1 Sol | 89.3 |
| GPT-5.6 Sol | 89.1 |
| GLM-5.3 Flash | 87.2 |
| GPT-5.6 Luna | 86.5 |