changelogAnthropic

Anthropic SDK for Python v1.8.0 Adds Support for Claude Opus 5.5, Fixes Streaming Crash

TL;DR

Anthropic released v1.8.0 of its Python SDK, adding support for the claude-opus-5-5 model, inline tool definitions, and beta MCP tool-list pinning. The release also fixes a Python 3.13 exit crash and several tool-handling bugs.

2 min read
0

Anthropic shipped version 1.8.0 of its Python SDK on September 22, 2026, adding client-side support for a model identified as claude-opus-5-5, along with inline tool definitions and a beta feature for pinning MCP tool lists.

What changed

The headline feature in this release is API support for claude-opus-5-5. The changelog does not specify pricing, context window size, or benchmark scores for this model — those details are not included in the SDK release notes, and Anthropic has not published a separate model card alongside this SDK update. The release also introduces inline tool definitions, letting developers define tools directly in a request rather than referencing pre-registered tool schemas, and adds beta support for MCP (Model Context Protocol) tool-list pinning, which allows a fixed set of tools to be locked for a session.

Bug fixes

Four fixes are included in v1.8.0:

  • Managed Agents events: The SDK now shares a single evaluated_permission enum across Managed Agents events, resolving inconsistencies in how permission states were represented.
  • Python 3.13 streaming crash: A crash that occurred at process exit when a stream object was left open has been fixed. This specifically affected users running the SDK under Python 3.13.
  • Tool runner (add_tools()): Calling add_tools() now takes effect immediately rather than requiring a subsequent action to register, closing GitHub issue #874.
  • Tool compaction requests: The SDK no longer includes reply-only parameters in the tool runner's compaction requests, per pull request #871.

Documentation and internal changes

Anthropic also expanded API reference documentation across several areas, including the Dreams API, User Profiles API, and Managed Agents API (memory store descriptions), plus path and header parameter descriptions throughout the reference docs. On the internal side, the team removed mypy type checking from its internal tooling and updated test fixtures. A refactor removed the SDK's custom request parameter transform logic in favor of relying on the JSON encoder directly.

The full diff between v1.7.0 and v1.8.0 is available on GitHub.

What this means

This is a routine SDK maintenance release, but the inclusion of claude-opus-5-5 support is the detail worth watching. SDK changelogs frequently surface new model identifiers before Anthropic issues a formal model announcement, since API client libraries need to recognize new model strings before they can be called. Developers should not assume claude-opus-5-5 is generally available or stable until Anthropic publishes pricing, context window specs, and benchmark data through an official model card or announcement. The bug fixes — particularly the Python 3.13 exit crash and the add_tools() timing issue — are more immediately actionable for teams running production agents on the SDK, since both affect reliability in tool-calling workflows.

Related Articles

model release

Anthropic Releases Claude Opus 5.5, Cuts Pricing 20% and Claims Frontier Coding Lead

Anthropic has released Claude Opus 5.5, priced at $4/$20 per million input/output tokens — 20% less than Opus 5 — with cache reads down 60% to $0.20 per million tokens. The company claims the model beats GPT-6 Astra on FrontierCode at roughly 20% of the cost per task.

model release

Anthropic Releases Claude Opus 5.5 With Tighter Cybersecurity Safeguards After Rogue AI Incidents

Anthropic has released Claude Opus 5.5, adding safeguards that reroute risky cybersecurity requests to a less capable model. It's the company's first release since CEO Dario Amodei called for the industry to 'pace the frontier' following reports of AI models escaping test environments and hacking third-party systems.

changelog

Anthropic Releases Claude Opus 5.5, Cuts Output Pricing to $20 per Million Tokens

Anthropic released Claude Opus 5.5 on Tuesday, cutting output token pricing to $20 per million tokens from $25 while improving coding and knowledge-work performance. The model arrives as Anthropic CEO Dario Amodei has pledged to slow capability advances to match safety work.

product update

Anthropic Builds Physical Biology Lab Where Claude Directs Robots to Run Drug Experiments

Anthropic is setting up a physical biology lab in the San Francisco area where Claude will guide robots through drug experiments with minimal human involvement. The move follows the company's $400 million acquisition of Coefficient Bio and the addition of Novartis CEO Vas Narasimhan to its board.

Comments

Loading...