Switch models by one parameter & compare free vs paid

USE 0 - 20 min

Same request, three models — change one string

The single most useful habit OpenRouter unlocks is swapping models without touching anything else. The request body stays identical; only the model field changes. That makes it trivial to send one prompt to several models and see how they differ — and to feel the difference between a free model and a paid flagship on the same task.

1 Take a working request from an earlier lesson (curl, Python, or your n8n node).
2 Run it once with "model": "google/gemma-4-31b-it:free" and keep the answer.
3 Change only the model field to a second free model, "nvidia/nemotron-3-super-120b-a12b:free", and run the same prompt again.
4 Change model once more to a paid flagship — "anthropic/claude-fable-5" — and run the identical prompt a third time. (This one is billed at the provider's rate: $0.00001 per prompt token, $0.00005 per completion token, with no OpenRouter markup.)
5 Lay the three answers side by side. Note where they agree, where the paid model is sharper, and where a free model was good enough.

✓

You have three answers to one prompt from three different models, produced by changing only the `model` string, and you can name one concrete difference between the free and paid responses.

UNDERSTAND 20 - 40 min

Pass-through pricing, and what 'free vs paid' really costs

Now that you have felt free and paid side by side, look at how OpenRouter prices them — because the pricing model is unusually simple and it changes how you choose.

OpenRouter homepage analytics graphic showing two overlaid panels — a Throughput line chart and a Latency chart with three coloured trend lines — on a dark background. — OpenRouter surfaces throughput and latency per model, so when you switch the `model` parameter you can also see what you trade in speed and responsiveness — not just quality. Source: https://openrouter.ai

Key concept

OpenRouter does not mark inference up. Per the live FAQ, verbatim: "We pass through the pricing of the underlying model providers without any markup, so you pay the same rate as you would directly with the provider." So 'free vs paid' is a real capability/cost trade, not a gateway tax: `:free` models cost $0 but, per the OpenRouter docs, "have low rate limits", while a paid model like `anthropic/claude-fable-5` is billed at the provider's exact per-token rate. Because switching is one parameter, you can prototype on free models and promote to a paid one only where the quality difference earns the cost.

?For the prompt you tested, was the paid model's answer worth its per-token cost — or was a free model good enough?
?Given pass-through pricing, what is the advantage of going through OpenRouter at all rather than calling the provider directly?
?A free model has low rate limits. For which of your tasks would that matter, and for which would it not?

BUILD 40 - 60 min

Build a model-comparison note for a real task

Turn the side-by-side habit into a decision you can defend — the kind of note you'd share before choosing a model for a pipeline.

Your task

Pick one real task from your work, run it through one free and one paid model by changing only the `model` parameter, and write a short verdict.

1 Choose a genuine task (a summary, a classification, an extraction) you actually need.
2 Run it on a :free model and on anthropic/claude-fable-5, keeping the rest of the request identical.
3 For each answer, mark one strength and one weakness.
4 Write a two-sentence verdict: which model you'd use for this task and whether the paid quality justified the per-token cost.

Deliverable

The prompt, both answers, and a two-sentence free-vs-paid verdict naming the model you'd ship for this task.