docs(bedrock): document the native Responses API route for OpenAI models

[mateo-berri]

TLDR

Adds a section to the Bedrock provider page for the native Responses API route that BerriAI/litellm#42767 shipped on 2026-09-23. Until now bedrock.md described OpenAI models on Bedrock only through the Converse bridge, so a reader had no way to learn that /v1/responses on bedrock/us.openai.* and bedrock/global.openai.* now goes straight to https://bedrock-runtime.{region}.amazonaws.com/openai/v1/responses, that the opt-in is the cost map's supported_endpoints: ["/v1/responses"], which models carry it today, or how the route differs from the bridge (prompt_cache_key works, background and web_search are dropped with a warning, remote image URLs are inlined)

Changes

One new ## OpenAI models on the native Responses API section in docs/providers/bedrock.md, placed after the OpenAI GPT OSS section and before TwelveLabs Pegasus, with an SDK and a proxy tab. Every claim in it comes from the merged PR's description and from litellm/llms/bedrock/responses/transformation.py on main. Nothing else on the page changes, so this does not overlap the pending chat-completions docs PR #1583, which waits on BerriAI/litellm#40775

---

Note

Low Risk
Documentation-only change to the Bedrock provider page; no runtime or configuration behavior is modified in this PR.

Overview
Adds OpenAI models on the native Responses API to docs/providers/bedrock.md, between the GPT OSS section and TwelveLabs Pegasus.

The new section explains that opt-in models (supported_endpoints: ["/v1/responses"] in the cost map, overridable via model_info / register_model) send /v1/responses directly to Bedrock’s openai/v1/responses endpoint instead of the Converse chat-completions bridge—so parameters like prompt_cache_key work and cached tokens show in usage. It lists which us. / global. profiles use the native route vs bridge-only models (e.g. GPT-OSS), notes unchanged auth/region/cost behavior and custom aws_bedrock_runtime_endpoint handling, and documents route-specific behavior (background / web_search dropped with warnings, remote images inlined, file_search emulated). SDK and proxy tabs include examples using litellm.responses and curl to /v1/responses.

^{Reviewed by Cursor Bugbot for commit 7f5db62. Bugbot is set up for automated code reviews on this repo. Configure here.}

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
litellm	Ready	Preview	Sep 24, 2026 9:26pm UTC

[mateo-berri]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 7f5db62. Configure here.}