Republishing Our Fable Customer Checks

Research evidence excerpt

Republishing Our Fable Customer Checks

Equity Research

Industry Update — July 1, 2026

U.S. Software

Our Call Ryan MacWilliams, CFA

We spoke with 2 engineering leaders at AI startups who briefly tested Fable 5 and Equity Analyst | Wells Fargo Securities, LLC

Ryan.MacWilliams@wellsfargo.com | 212-214-1192

reported a step function improvement in AI performance capability. Both noted the

Michael Turrin, CFAhigher ROI was well worth the higher token cost, which can easily surpass $100/hour.

Equity Analyst | Wells Fargo Securities, LLC

Michael.Turrin@wellsfargo.com | 628-629-7565

Fable 5 is much better at completing higher-ROI, long-running tasks on its first Ken Gawrelski

try, that previously required multiple turns of human + AI review. Checks report more Equity Analyst | Wells Fargo Securities, LLC

instances of Fable working autonomously (like a reliable coworker), while prior models Ken.Gawrelski@wellsfargo.com | 617-210-3707

needed more direct supervision. Zeeshan Rauf

Associate Equity Analyst | Wells Fargo Securities, LLC

• More human-like performance comes at a more human-like cost, and Fable-like Zeeshan.Rauf@wellsfargo.com | 212-214-1120

models could reshape conversation around which models are deployed where, for Chris Brazeau

what use case, and token spend estimates. Associate Equity Analyst | Wells Fargo Securities, LLC

Chris.Brazeau@wellsfargo.com | 929-767-5658

Figure 1 - Fable 5's performance on Agent Arena shows strong improvements vs. other Cyrus Motakef

leading AI models Associate Equity Analyst | Wells Fargo Securities, LLC Cyrus.Motakef@wellsfargo.com | 212-214-3334

Source: Arena.ai, Hugging Face, and Wells Fargo Securities, LLC

Note: Agent Arena (link) is a dynamic ranking of models based on how well they orchestrate tools for