anthropics / anthropics/anthropic-sdk-python

Bedrock: support low-latency configuration

Open
#971 0 comments 0 reactions 0 assignees View on GitHub
bedrock
Dominant language
Python
Stars
3.9k
Forks
853
Avg merge
1d 11h
Merged PRs (30d)
10

Description

Currently, `[Async]AnthropicBedrock` doesn't provide a way to enable low-latency mode.
It also appears that this configuration isn't even accessible via the Bedrock Anthropic HTTP API.

So the only option for the Antropic clients to use this low-latency model is to use the Converse API provided by AWS Becrock: https://docs.aws.amazon.com/bedrock/latest/userguide/latency-optimized-inference.html

This is very inconvenient, since the Converse API unifies APIs for multiple models and, therefore, it lacks support for exclusive features of particular models _(e.g. `tool_choice=none` is missing from the Converse API)_.

Exposing low-latency configuration in `[Async]AnthropicBedrock` would be great.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.