How does the AI model comparison calculate cost?
It multiplies your entered input tokens, output tokens, and monthly requests by the active source-tracked pricing profile for each selected model.
Can I compare coding-tool subscriptions and API models?
Yes, but they are separate tabs because API models use token workload pricing while coding tools use plan and product-fit fields.
Does the comparison include cached-input pricing?
Only when a selected model has tracked cached-input pricing and you enter a cache-hit percentage. The default is 0% cache hits.
Why are some fields unavailable?
StackLens leaves fields unavailable when the verified dataset does not contain enough source-tracked information to calculate or compare them safely.
Does lower cost mean better model quality?
No. Lower cost is only one signal. Test quality, latency, retries, output length, and workflow fit before routing production traffic.