[ // 1 credit = X tokens (sum of input + output). // ceil(total_tokens / tokens_per_credit) is debited per call. 'tokens_per_credit' => 150, ], 'image' => [ // Credits charged per image generated, keyed by model id. // Falls back to 'default' if the model is not listed. 'default' => 50, 'gpt-image-1.5' => 50, 'gpt-image-1.5-hd' => 100, // gpt-image-2 is currently called at quality=low (~$0.01/image at 1024². // If we ever expose medium/high to users, add gpt-image-2-medium / -hd. 'gpt-image-2' => 15, 'flux-pro' => 30, ], 'video' => [ 'default' => 500, ], ];