Anthropic Python SDK 1.9.0 Adds Reference to Unreleased 'claude-sonnet-5-5' Model ID
Anthropic's anthropic-sdk-python v1.9.0 release adds a reference to an unannounced 'claude-sonnet-5-5' model ID, a new between_tools thinking type, and the ability to run tool calls while a reply streams. No pricing, context window, or benchmark data for the model has been disclosed.
Anthropic SDK v1.9.0 Ships With Unannounced Model Reference
Anthropic released version 1.9.0 of its anthropic-sdk-python library on September 28, 2026, adding a code-level reference to a model ID called claude-sonnet-5-5 — a name that has not been formally announced or documented by Anthropic elsewhere. The release also introduces a new between_tools thinking type and lets developers run tool calls while a reply is still streaming.
This is a software development kit update, not a model launch. No context window size, pricing, benchmark scores, or parameter count for claude-sonnet-5-5 accompanies this release. The string appears only as an addition to the SDK's list of known model IDs, alongside a separate chore that reorders how known model IDs are displayed in the SDK's type definitions.
What changed in v1.9.0
Features:
- A new
between_toolsthinking type, letting models reason between sequential tool calls - Addition of the
claude-sonnet-5-5model identifier to the SDK's type list - New parameters (
include_inherited,source) for workspace rate-limit objects - Typed event filters for the Managed Agents events list
- General availability of cache diagnostics on
MessageandMessageCreateParamsobjects - Optional execution of tool calls while a streamed reply is still being generated
Bug fixes:
- The client no longer sends a placeholder filename for unnamed file uploads
between_toolsthinking now degrades to disabled during fallback hops, fixing a helper-level bug (#952)stream()andparse()methods now correctly accept diagnostics data (#929)
Chores and documentation:
- Known model IDs are now listed first in the SDK's Model type definitions
- CI runner selection now varies by repository
- The beta header is no longer sent from
parse()and the tool runner - Documentation clarifies that
stream: truereturns the raw event stream - Managed Agents actor descriptions were made resource-neutral
- A research-preview notice was restored on the
Dreamtype
Full commit history is available in Anthropic's GitHub compare view.
What this means
The appearance of claude-sonnet-5-5 in the SDK's type definitions is the most notable detail here, but it stops well short of confirming a model release. Companies routinely stage model IDs in client libraries ahead of public announcements to reduce integration lag once a model does ship — this happened with earlier Claude versions too. Until Anthropic publishes specs, pricing, or a model card for claude-sonnet-5-5, treat the reference as a signal of an upcoming release, not evidence one has occurred.
The more concrete changes in this release — streaming tool execution and the between_tools thinking type — are developer-facing improvements to how Claude models handle multi-step tool use, letting agents act on partial output instead of waiting for a full response. These are meaningful for latency-sensitive agent applications, but they modify how existing models are accessed, not the models themselves.
Related Articles
Claude can now build HTML-in-Canvas sites, needs Chrome or Edge 157 or Claude desktop 5.11492
Claude can now build web pages that embed real HTML elements such as text and buttons inside canvas graphics and 3D effects. The feature works in Chrome and Edge from version 157 and in Claude's desktop apps from version 5.11492. Users must request HTML-in-Canvas explicitly in their prompt.
OpenAI Python SDK v3.56.0 adds gpt-live-1.1 to Live model choices
OpenAI's Python SDK v3.56.0, dated 2026-11-19, adds gpt-live-1.1 to the Live model choices. The release lists this as its only feature.
OpenAI Python SDK v3.53.0 adds GPT-6.2 Astra, Astra Law and Sol identifiers
OpenAI Python SDK v3.53.0, dated 2026-11-17, adds model identifiers for GPT-6.2 Astra, GPT-6.2 Astra Law and GPT-6.2 Sol. It also adds 2-hour and 8-hour Responses background windows.
Anthropic: Selective notes beat full logs for Claude agent memory, with 'Dreaming' fixing errors
Anthropic published research on how Claude retains knowledge across sessions using note files. Its tests show selective notes outperform exhaustive archives, and a post-session process called Dreaming can correct faulty notes. All results come from Anthropic's own tests.
Comments
Loading...