Anthropic SDK
AI & LLMsStandardized changelog data for anthropic-sdk.
Release History
v0.84.0Breaking3 featuresVersion 0.84.0 updates API serialization for arrays to use bracket notation and introduces new conversion helpers for MCP tools and prompts. The project documentation has also been rebranded to Claude SDK.
v0.83.01 featureThis release introduces top-level automatic caching control to the API settings. It also includes minor documentation updates for the mock server.
v0.82.01 fix1 featureThis release introduces fixes for shared UserLocation and error code types in the API, along with backward-compatibility aliases for previously removed nested UserLocation classes.
v0.81.01 featureThis release introduces new tool versions as top-level types within the API, enhancing tool usage capabilities.
v0.80.02 fixes1 featureVersion 0.80.0 introduces the new claude-sonnet-4-6 model and includes several bug fixes, such as correcting API specification errors.
v0.79.01 fix1 featureThis release introduces the ability to enable fast-mode for claude-opus-4-6 via the API and fixes a bug related to passing the speed parameter in count_tokens.
v0.78.01 featureVersion 0.78.0 introduces manual updates to the API endpoints. This release focuses on keeping the SDK synchronized with the latest API specifications.
v0.77.11 fixThis patch release fixes an issue where the structured output beta header was not being sent correctly when the format was omitted. It also removes an obsolete workflow.
v0.77.0Breaking2 fixes3 featuresThis release introduces support for Structured Outputs in the Messages API, migrating the configuration mechanism to use `output_config`. It also includes several bug fixes related to structured output headers and general client improvements.
v0.76.04 fixes3 featuresThis release introduces support for raw JSON schema in streaming messages, binary request streaming, and server-side tools. It also includes several bug fixes, notably loosening auth header validation and ensuring streams are closed.
v0.75.02 fixes5 featuresThis release introduces support for Claude Opus 4.5 along with advanced features including Computer Use v5, Autocompaction, and enhanced tool use capabilities.
v0.74.11 fixThis patch release fixes an issue with the beta header for structured outputs and updates model references in the examples.
v0.74.02 fixes1 featureThis release introduces the Foundry SDK and includes bug fixes for cross-platform path handling and memory examples.
v0.73.01 featureThis release introduces beta support for structured outputs in the Anthropic API.
v0.72.1Breaking3 fixesThis release drops support for Python 3.8 and includes bug fixes for stream handling, Pydantic v1 compatibility, and Python 3.14 support.
v0.72.01 featureThis release introduces the ability to clear thinking within context management in the API.
v0.71.11 fixThis patch release fixes non-functional socket options in the client and updates internal model deprecation statuses and dependencies.
v0.71.01 featureThis release introduces support for agent skills within the Anthropic API.
v0.70.03 featuresThis release introduces manual API updates and adds beta headers for context management and model context window exceeded features.
v0.69.02 featuresThis release introduces support for the Claude Sonnet 4.5 model and adds new context management features to the API.
v0.68.21 fixThis patch release fixes a bug where headers with default values were incorrectly being set to omit.
v0.68.1This maintenance release focuses on internal type system updates, specifically transitioning optional parameter types from NotGiven to Omit, and cleaning up project dependency configurations.
v0.68.01 featureThis release introduces new tool running helpers and includes internal maintenance updates, specifically regarding Pydantic dependency management and test suite improvements.
v0.67.02 fixes2 featuresThis release introduces support for the web_fetch_20250910 tool and focuses on internal improvements for Pydantic v3 compatibility and beta type definitions.
v0.66.01 featureThis release introduces support for including Documents within tool results in the Anthropic Python SDK.
v0.65.02 fixes2 featuresThis release introduces support for the code-execution-2025-08-26 tool and updates type hints across parameters to use SequenceNotStr instead of List[str].
v0.64.01 featureThis release promotes 1-hour TTL Cache Control to general availability and deprecates older Claude 3.5 Sonnet model versions.
v0.63.01 featureThis release introduces support for the context-1m-2025-08-07 beta and includes internal maintenance updates for breaking change detection and testing.
v0.62.01 featureThis release introduces support for search result content blocks within the API.
v0.61.04 featuresThis release introduces support for the Claude Opus 4-1 model, new text editor tools, and file upload capabilities.
v0.60.01 fix1 featureThis release adds missing beta methods to the Vertex AI integration and updates streaming error messages for better clarity on requirements.
v0.59.0Breaking2 fixes1 featureRelease 0.59.0 removes support for older deprecated models and improves response parsing for metadata and extra field types.
v0.58.2This is a maintenance release involving an internal version bump to v0.58.2.
v0.58.1This is a minor internal version bump release with no functional changes to the codebase.
v0.58.02 fixes2 featuresThis release introduces official support for Python 3.13 and includes bug fixes for GET request headers and nested union parsing.
v0.57.1This is a minor maintenance release that updates the BetaCitationSearchResultLocation API model and performs an internal version bump.
v0.57.02 fixes1 featureThis release introduces support for Search Result Content Blocks and improves error messaging for network timeouts and stream processing.
v0.56.03 fixes2 featuresThis release introduces automatic AWS Region inference for Bedrock and support for global region endpoints in Vertex AI, alongside several CI and test suite improvements.
v0.55.04 fixes2 featuresThis release introduces support for aiohttp and includes several API updates, alongside bug fixes for binary response parsing and test suite improvements.
v0.54.02 fixes1 featureThis release introduces support for fine-grained tool streaming and resolves a conflict between default transport and proxy settings in the httpx client.
v0.53.02 fixes2 featuresThis release introduces support for the text_editor_20250429 tool and a new follow_redirects option, while deprecating BetaBase64PDFBlock in favor of BetaRequestDocumentBlock.
v0.52.22 fixesThis patch release fixes an issue with server_tool_use input tracking and transitions the internal package management from rye to uv.
v0.52.11 fix1 featureThis patch release focuses on fixing a broken asset in the examples and providing new documentation on MCP server authorization.
v0.52.02 fixes4 featuresThis release introduces support for Claude 4 models along with major new capabilities including a Files API, code execution tools, and MCP connectors.
v0.51.02 fixes1 featureThis release introduces web search capabilities to the Claude API and includes bug fixes for Pydantic v1 integration and socket handling.
v0.50.0Breaking7 fixes2 featuresRelease 0.50.0 introduces schema changes for ContentBlockDelta events, improves performance by optimizing hot paths, and bumps the minimum httpx dependency to 0.25.0.
v0.49.01 featureThis release introduces the ability to disable tool calls in the API and includes updates to client docstrings and documentation URLs.
v0.48.01 featureThis release introduces support for URL source blocks for images and PDFs in the API, along with new documentation examples for thinking.
v0.47.21 fixThis release fixes a bug in the beta streaming interface to include thinking blocks and includes internal Pydantic configuration improvements.
v0.47.11 fixThis patch release includes internal specification updates and improves the error messaging for large max_tokens values.
v0.47.01 fix3 featuresThis release introduces support for the Claude 3.7 model and its new thinking capabilities, along with improved error handling and more flexible request body parameters.
v0.46.03 fixes3 featuresRelease 0.46.0 introduces a .close() method for JSONL handling, optimized pagination, and fixes for async execution on non-asyncio runtimes.
v0.45.21 fixThis patch release fixes a deserialization error encountered during message streaming.
v0.45.11 fixThis patch release fixes a bug in the streaming functionality to ensure citations are correctly accumulated.
v0.45.02 featuresThis release introduces support for citations in the API and adds a new results endpoint to the client.
v0.44.02 fixes1 featureThis release introduces a request_id getter for streaming responses and deprecates several older models, alongside internal test improvements and style updates.
v0.43.11 fix2 featuresThis patch release focuses on internal improvements, including better error messaging for streaming, enhanced type metadata, and documentation corrections.
v0.43.0Breaking3 fixes2 featuresThis release introduces message batch deletion and beta streaming helpers, while cleaning up deprecated Vertex AI client options and improving stream handling internals.
Common Errors
BadRequestError14 reportsBadRequestError in the anthropic-sdk usually arises from sending malformed requests to the Claude API, such as including unexpected or incorrectly formatted data fields like extra headers. To resolve this, carefully inspect your request payload, especially custom headers, output format specifications, and any extra fields, ensuring they strictly adhere to the API's requirements and documented schema; remove any extraneous or incorrect elements that are not explicitly supported.
NotFoundError4 reportsThe "NotFoundError" in anthropic-sdk often arises from incorrect package installation or version conflicts, preventing the program from locating necessary modules or attributes. Ensure that the anthropic-sdk is correctly installed (`pip install anthropic --upgrade`) and that its version aligns with the code's requirements, checking for potential conflicts with other installed packages. Downgrading or upgrading anthropic-sdk to a compatible version might also be necessary.
RateLimitError3 reportsRateLimitError in anthropic-sdk usually means you've exceeded your API usage limits. To fix this, check your Anthropic account to understand your rate limits and usage, then implement retry logic with exponential backoff in your code to gracefully handle rate limiting.
InternalServerError3 reportsInternalServerError in anthropic-sdk often arises from backend issues on Anthropic's servers or exceeding rate limits. Ensure your API key is valid, check Anthropic's status page for service disruptions, and implement retry logic with exponential backoff to handle temporary server unavailability or rate limiting. If using beta/preview features, upgrade to the latest sdk version or check for correct usage as potential code changes can cause unexpected errors.
PermissionDeniedError2 reportsThe PermissionDeniedError in anthropic-sdk usually stems from insufficient permissions to access the required files or resources, often during installation. Try running the installation with elevated privileges (e.g., using `sudo pip install anthropic` or installing within a virtual environment) to grant the necessary access rights. You may also need to check and correct the permissions of the relevant directories involved in the installation process.
UnprocessableEntityError2 reports"UnprocessableEntityError" in anthropic-sdk often arises from incorrect or missing data in your API request body like misspelled fields or wrong data types. Carefully review your request payload for compliance with the expected schema in the Anthropic API documentation, paying special attention to required fields and allowed values. Ensure your code accurately serializes the data before sending it in the request.
Related AI & LLMs Packages
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.
Get up and running with OpenAI gpt-oss, DeepSeek-R1, Gemma 3 and other models.
🦜🔗 The platform for reliable agents.
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
LLM inference in C/C++
GPT4All: Run Local LLMs on Any Device. Open-source and available for commercial use.
Subscribe to Updates
Get notified when new versions are released