AWS bedrock documentation change
Summary
Added documentation for new Claude Mythos 5 and Claude Fable 5 models, updated supported thinking types, and introduced new 'Connector text summarization' beta feature
Security assessment
Changes describe new model capabilities and response handling patterns without addressing vulnerabilities or security controls. The connector text summarization focuses on response formatting rather than security mechanisms.
Diff
diff --git a/bedrock/latest/userguide/claude-messages-adaptive-thinking.md b/bedrock/latest/userguide/claude-messages-adaptive-thinking.md index 64f0badb9..84374de05 100644 --- a//bedrock/latest/userguide/claude-messages-adaptive-thinking.md +++ b//bedrock/latest/userguide/claude-messages-adaptive-thinking.md @@ -7 +7 @@ -How adaptive thinking worksAdaptive thinking with the effort parameterUsing adaptive thinking with the Converse APIPrompt cachingTuning thinking behavior +How adaptive thinking worksAdaptive thinking with the effort parameterUsing adaptive thinking with the Converse APIPrompt cachingTuning thinking behaviorConnector text summarization (beta) @@ -16,0 +17,2 @@ Model | Model ID +Claude Mythos 5 | `anthropic.claude-mythos-5` +Claude Fable 5 | `anthropic.claude-fable-5` @@ -24 +26 @@ Claude Sonnet 4.6 | `anthropic.claude-sonnet-4-6` -Claude Opus 4.7 and Claude Mythos Preview _only_ support adaptive thinking. Manual extended thinking (`thinking.type: "enabled"` with `budget_tokens`) is not supported on these models and will return a 400 error. +Claude Mythos 5, Claude Fable 5, Claude Opus 4.7, and Claude Mythos Preview _only_ support adaptive thinking. Manual extended thinking (`thinking.type: "enabled"` with `budget_tokens`) and disabled thinking (`thinking.type: "disabled"`) are not supported on these models and will return a 400 error. Use `thinking.type: "adaptive"` with `output_config.effort` to control thinking behavior. @@ -225,0 +228,17 @@ Steering Claude to think less often may reduce quality on tasks that benefit fro +## Connector text summarization (beta) + +On Claude Fable 5, text that the model emits between tool calls (sometimes called "connector text" — for example, "Let me check that file next...") is summarized server-side and returned as a thinking block rather than a plain text content block. The thinking block uses the same shape as any other thinking block (empty text with a signature under the default `omitted` display). + +**Customer impact:** + + * **Response shape:** Tool-use responses from Claude Fable 5 may contain additional thinking blocks where previous models emitted plain text between `tool_use` blocks. There is no new content block type. Final assistant answers (after all tool use is complete) are unaffected and remain plain text. + + * **Multi-turn handling:** Pass these thinking blocks back unchanged in multi-turn conversations — the same handling as protected thinking (signature validated on passback; silently stripped if sent to a different model). + + * **Scope:** Connector summarization applies only after a `tool_result` exists in the conversation. Narration before the first tool call in a fresh conversation remains plain text. Short text segments may pass through as plain text without summarization. + + + + +This feature is enabled server-side for Claude Fable 5. There is no customer opt-in or opt-out. +