TradePerformance

What is the difference between the free and paid models?

The free models are genuinely capable at writing and summarising and genuinely weaker at reasoning, at seeing images reliably, and at calling tools. The paid models are better at all three, and the gap shows up most on the features that chain several steps together.

The app's own model list records exactly which model can do what, and it is worth reading it that way rather than as "good" versus "bad".

Three capabilities matter here:

On the free tier, no single model has all three. Groq's default, Llama 3.3 70B, is strong at text but cannot see images and its tool calls fail often enough that the app routes around it. Llama 4 Scout can see and can call tools but is a smaller model. On OpenRouter, Gemini 2.0 Flash covers vision and tools; DeepSeek R1 Distill covers reasoning and neither of the others.

That is precisely why smart routing exists — it picks the free model that can actually do the task in front of it rather than failing on the one you selected. See smart model routing.

On the paid tier, every listed model does all three. The choice there is about depth and price, not capability.

the routing rules, which spell out how free and paid providers are treated differently
the routing rules, which spell out how free and paid providers are treated differently