[Feature]: pricing for 1h cache writes with long-context models

[nuernber]

The Feature

It looks like LiteLLM already has cache_creation_input_token_cost_above_1hr, but it doesn't have a notion for cache_creation_input_token_cost_above_1hr_above_200k_tokens. That is, from the AWS Bedrock pricing page (https://aws.amazon.com/bedrock/pricing/), the following red circled price isn't getting tracked right now:

Motivation, pitch

Pricing is not working correctly right now for this situation

What part of LiteLLM is this about?

Proxy

LiteLLM is hiring a founding backend engineer, are you interested in joining us and shipping to all our users?

No

Twitter / LinkedIn details

No response

[DmitriyAlergant]

To double down here, the x2 x2 cost multiplicator of (long context Sonnet) x (1h caching) was always Anthropic's pricing policy not limited to Bedrock. It's the same in direct Anthropic API. I guess not many users are using 1hr caching.

[nuernber]

Per https://platform.claude.com/docs/en/build-with-claude/context-windows#1-m-token-context-window, it sounds like those in "usage tier 4" only have access to it.