Meta models
2 tracked models from Meta, priced at the standard API tier. Llama 4 Maverick is the cheapest to run; Llama 3.3 70B sits at the top of the lineup. Quality scores per task are benchmark-informed estimates — validate on your own workload before you migrate.
Models
2
Blended /1M
$0.41–$0.57
Max context
1M
Open weights
2 of 2
ModelInput /1MOutput /1MBlended /1M