Model Multipliers
Not all LLM models cost the same to run. Quarterback uses multipliers to normalize usage across models so billing stays fair and transparent.
How Multipliers Work
Each model has a multiplier that determines how many compute hours it consumes relative to the baseline. The baseline (1.0x) is set to a standard-tier model.
| Model Tier | Example | Multiplier |
|---|---|---|
| Standard | Claude Sonnet | 1.0x |
| Advanced | Claude Opus | 3.0x |
| Fast | Claude Haiku | 0.25x |
For example, if a turn takes 60 seconds of wall-clock time:
- With a 1.0x model: 60 seconds consumed
- With a 3.0x model: 180 seconds consumed
- With a 0.25x model: 15 seconds consumed
Choosing Models
Your organization admin can configure which models are available at the org level. If cost is a concern, restricting to standard-tier models keeps your multiplier at 1.0x.
Model availability is configured in Org Settings > Models.
Impact on Billing
Multipliers only affect how quickly you consume your included hours (or credits). They don’t change the price of your plan or credits.
Example: An org with 100 included hours per month using exclusively a 3.0x model would get ~33 effective hours of usage. The same org using a 0.25x model would get ~400 effective hours.
Checking Model Multipliers
The current multiplier for each model is visible in:
- Org Settings > Models — shows all allowed models with their multipliers
- Session stats — the right sidebar shows the model and multiplier for the current session