Are you using Claude, GPT-4o, or Llama for your production agent, and what made you choose?

Genuine question because the answer keeps changing.

Six months ago GPT-4o was the default for most teams. Then Claude 3.5 Sonnet started winning reasoning-heavy use cases. Now Llama 3 is showing up in production for cost-sensitive or on-prem deployments.

What's your current production model, and what was the actual deciding factor? Quality, cost, latency, compliance, or something else?

Bonus question, are you running one model across all agents or routing per use case?