Bug description
In the AWS BedrockAnthropic3ChatModel sync call the token usage metrics is not returned.
Also bedrock return other very important metric X-Amzn-Bedrock-Invocation-Latency that would be good to include in the spring ai response model
Environment
Last M3 version
Steps to reproduce
Make a simple call with the BedrockAnthropic3ChatModel
Expected behavior
Have all response metrics that bedrock returns
Thanks in advance