Returns

• BetaMessage object

  • type: "message"

    Object type.

    For Messages, this is always "message".

    default: message

  • id: string

    Unique object identifier.

    The format and length of IDs may change over time.

  • container: BetaContainer or null

    Information about the container used in the request (for the code execution tool)

    • id: string

      Identifier for the container used in this request

    • expires_at: string

      The time at which the container will expire.

      format: date-time

    • skills: array of BetaContainerSkill or null

      Skills loaded in the container

      • type: "anthropic" or "custom"

        Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)

        • "anthropic"

        • "custom"

      • skill_id: string

        Skill ID

        maxLength: 64, minLength: 1

      • version: string

        The resolved version: a skill version ID for custom skills.

        maxLength: 64, minLength: 1

  • content: array of BetaContentBlock

    Content generated by the model.

    This is an array of content blocks, each of which has a type that determines its shape.

    Example:

    [\{"type": "text", "text": "Hi, I'm Claude."\}]

    If the request input messages ended with an assistant turn, then the response content will continue directly from that last turn. You can use this to constrain the model's output.

    For example, if the input messages were:

    [
      \{"role": "user", "content": "What's the Greek name for Sun? (A) Sol (B) Helios (C) Sun"\},
      \{"role": "assistant", "content": "The best answer is ("\}
    ]

    Then the response content might be:

    [\{"type": "text", "text": "B)"\}]

    • BetaTextBlock object

      • type: "text"

        default: text

      • citations: array of BetaTextCitation or null

        Citations supporting the text block.

        The type of citation returned will depend on the type of document being cited. Citing a PDF results in page_location, plain text results in char_location, and content document results in content_block_location.

        • BetaCitationCharLocation object

          • type: "char_location"

            default: char_location

          • cited_text: string

          • document_index: number

            minimum: 0

          • document_title: string or null

          • end_char_index: number

          • file_id: string or null

          • start_char_index: number

            minimum: 0

        • BetaCitationPageLocation object

          • type: "page_location"

            default: page_location

          • cited_text: string

          • document_index: number

            minimum: 0

          • document_title: string or null

          • end_page_number: number

          • file_id: string or null

          • start_page_number: number

            minimum: 1

        • BetaCitationContentBlockLocation object

          • type: "content_block_location"

            default: content_block_location

          • cited_text: string

            The full text of the cited block range, concatenated.

            Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.

          • document_index: number

            minimum: 0

          • document_title: string or null

          • end_block_index: number

            Exclusive 0-based end index of the cited block range in the source's content array.

            Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.

          • file_id: string or null

          • start_block_index: number

            0-based index of the first cited block in the source's content array.

            minimum: 0

        • BetaCitationsWebSearchResultLocation object

          • type: "web_search_result_location"

            default: web_search_result_location

          • cited_text: string

          • encrypted_index: string

          • title: string or null

            maxLength: 512

          • url: string

        • BetaCitationSearchResultLocation object

          • type: "search_result_location"

            default: search_result_location

          • cited_text: string

            The full text of the cited block range, concatenated.

            Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.

          • end_block_index: number

            Exclusive 0-based end index of the cited block range in the source's content array.

            Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.

          • search_result_index: number

            0-based index of the cited search result among all search_result content blocks in the request, in the order they appear across messages and tool results.

            Counted separately from document_index; server-side web search results are not included in this count.

            minimum: 0

          • source: string

          • start_block_index: number

            0-based index of the first cited block in the source's content array.

            minimum: 0

          • title: string or null

      • text: string

        minLength: 0

    • BetaThinkingBlock object

      • type: "thinking"

        default: thinking

      • signature: string

        A value used to verify that this thinking block was generated by Claude when it is passed back to the API.

        This is an opaque field and should not be interpreted or parsed. When passing thinking blocks back to the API (required when using tools with extended thinking), pass them back exactly as received, with this field intact.

        See extended thinking for details.

      • thinking: string

        The text of Claude's thinking process for this block.

    • BetaRedactedThinkingBlock object

      • type: "redacted_thinking"

        default: redacted_thinking

      • data: string

        The contents of this redacted thinking block, returned when portions of the model's thinking were safety-redacted. This field is opaque and encrypted, with no readable content.

        Pass redacted_thinking blocks back to the API unchanged when continuing a multi-turn conversation.

        See extended thinking for details.

    • BetaToolUseBlock object

      • type: "tool_use"

        default: tool_use

      • id: string

        pattern: ^[a-zA-Z0-9_-]+$

      • input: map[unknown]

      • name: string

        minLength: 1

      • caller: optional BetaDirectCaller or BetaServerToolCaller or BetaServerToolCaller20260120

        Tool invocation directly from the model.

        • BetaDirectCaller object

          Tool invocation directly from the model.

          • type: "direct"

        • BetaServerToolCaller object

          Tool invocation generated by a server-side tool.

          • type: "code_execution_20250825"

          • tool_id: string

            pattern: ^srvtoolu_[a-zA-Z0-9_]+$

        • BetaServerToolCaller20260120 object

          • type: "code_execution_20260120"

          • tool_id: string

            pattern: ^srvtoolu_[a-zA-Z0-9_]+$

      • toolset_name: optional string or null

        For a toolset member tool_use, the toolset family.

        maxLength: 64, minLength: 1, pattern: ^[a-zA-Z0-9_-]+$

    • BetaServerToolUseBlock object

      • type: "server_tool_use"

        default: server_tool_use

      • id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

      • input: map[unknown]

      • name: "advisor" or "web_search" or "web_fetch" or 5 more

        • "advisor"

        • "web_search"

        • "web_fetch"

        • "code_execution"

        • "bash_code_execution"

        • "text_editor_code_execution"

        • "tool_search_tool_regex"

        • "tool_search_tool_bm25"

      • caller: optional BetaDirectCaller or BetaServerToolCaller or BetaServerToolCaller20260120

        Tool invocation directly from the model.

        • BetaDirectCaller object

          Tool invocation directly from the model.

        • BetaServerToolCaller object

          Tool invocation generated by a server-side tool.

        • BetaServerToolCaller20260120 object

    • BetaWebSearchToolResultBlock object

      • type: "web_search_tool_result"

        default: web_search_tool_result

      • content: BetaWebSearchToolResultBlockContent

        • BetaWebSearchToolResultError object

          • type: "web_search_tool_result_error"

            default: web_search_tool_result_error

          • error_code: BetaWebSearchToolResultErrorCode

            • "invalid_tool_input"

            • "unavailable"

            • "max_uses_exceeded"

            • "too_many_requests"

            • "query_too_long"

            • "request_too_large"

        • array of BetaWebSearchResultBlock

          • type: "web_search_result"

            default: web_search_result

          • encrypted_content: string

          • page_age: string or null

          • title: string

          • url: string

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

      • caller: optional BetaDirectCaller or BetaServerToolCaller or BetaServerToolCaller20260120

        Tool invocation directly from the model.

        • BetaDirectCaller object

          Tool invocation directly from the model.

        • BetaServerToolCaller object

          Tool invocation generated by a server-side tool.

        • BetaServerToolCaller20260120 object

    • BetaWebFetchToolResultBlock object

      • type: "web_fetch_tool_result"

        default: web_fetch_tool_result

      • content: BetaWebFetchToolResultErrorBlock or BetaWebFetchBlock

        • BetaWebFetchToolResultErrorBlock object

          • type: "web_fetch_tool_result_error"

            default: web_fetch_tool_result_error

          • error_code: BetaWebFetchToolResultErrorCode

            • "invalid_tool_input"

            • "url_too_long"

            • "url_not_allowed"

            • "url_not_in_prior_context"

            • "url_not_accessible"

            • "unsupported_content_type"

            • "too_many_requests"

            • "max_uses_exceeded"

            • "unavailable"

            • "content_too_large"

        • BetaWebFetchBlock object

          • type: "web_fetch_result"

            default: web_fetch_result

          • content: BetaDocumentBlock

            • type: "document"

              default: document

            • citations: BetaCitationConfig or null

              Citation configuration for the document

              • enabled: boolean

                default: false

            • source: BetaBase64PDFSource or BetaPlainTextSource

              • BetaBase64PDFSource object

                • type: "base64"

                • data: string

                  format: byte

                • media_type: "application/pdf"

              • BetaPlainTextSource object

                • type: "text"

                • data: string

                • media_type: "text/plain"

            • title: string or null

              The title of the document

          • retrieved_at: string or null

            ISO 8601 timestamp when the content was retrieved

          • url: string

            Fetched content URL

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

      • caller: optional BetaDirectCaller or BetaServerToolCaller or BetaServerToolCaller20260120

        Tool invocation directly from the model.

        • BetaDirectCaller object

          Tool invocation directly from the model.

        • BetaServerToolCaller object

          Tool invocation generated by a server-side tool.

        • BetaServerToolCaller20260120 object

    • BetaAdvisorToolResultBlock object

      • type: "advisor_tool_result"

        default: advisor_tool_result

      • content: BetaAdvisorToolResultError or BetaAdvisorResultBlock or BetaAdvisorRedactedResultBlock

        • BetaAdvisorToolResultError object

          • type: "advisor_tool_result_error"

            default: advisor_tool_result_error

          • error_code: "max_uses_exceeded" or "prompt_too_long" or "too_many_requests" or 4 more

            • "max_uses_exceeded"

            • "prompt_too_long"

            • "too_many_requests"

            • "overloaded"

            • "unavailable"

            • "execution_time_exceeded"

            • "model_not_found"

        • BetaAdvisorResultBlock object

          • type: "advisor_result"

            default: advisor_result

          • stop_reason: string or null

            The advisor sub-inference's stop reason (same values as the top-level message stop_reason). max_tokens indicates the advisor's output was truncated at the tool's max_tokens value or the advisor model's policy cap.

          • text: string

        • BetaAdvisorRedactedResultBlock object

          • type: "advisor_redacted_result"

            default: advisor_redacted_result

          • encrypted_content: string

            Opaque blob containing the advisor's output. Round-trip verbatim; do not inspect or modify.

          • stop_reason: string or null

            The advisor sub-inference's stop reason (same values as the top-level message stop_reason).

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

    • BetaCodeExecutionToolResultBlock object

      • type: "code_execution_tool_result"

        default: code_execution_tool_result

      • content: BetaCodeExecutionToolResultBlockContent

        Code execution result with encrypted stdout for PFC + web_search results.

        • BetaCodeExecutionToolResultError object

          • type: "code_execution_tool_result_error"

            default: code_execution_tool_result_error

          • error_code: BetaCodeExecutionToolResultErrorCode

            • "invalid_tool_input"

            • "unavailable"

            • "too_many_requests"

            • "execution_time_exceeded"

        • BetaCodeExecutionResultBlock object

          • type: "code_execution_result"

            default: code_execution_result

          • content: array of BetaCodeExecutionOutputBlock

            • type: "code_execution_output"

              default: code_execution_output

            • file_id: string

          • return_code: number

          • stderr: string

          • stdout: string

        • BetaEncryptedCodeExecutionResultBlock object

          Code execution result with encrypted stdout for PFC + web_search results.

          • type: "encrypted_code_execution_result"

            default: encrypted_code_execution_result

          • content: array of BetaCodeExecutionOutputBlock

            • type: "code_execution_output"

              default: code_execution_output

            • file_id: string

          • encrypted_stdout: string

          • return_code: number

          • stderr: string

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

    • BetaBashCodeExecutionToolResultBlock object

      • type: "bash_code_execution_tool_result"

        default: bash_code_execution_tool_result

      • content: BetaBashCodeExecutionToolResultError or BetaBashCodeExecutionResultBlock

        • BetaBashCodeExecutionToolResultError object

          • type: "bash_code_execution_tool_result_error"

            default: bash_code_execution_tool_result_error

          • error_code: "invalid_tool_input" or "unavailable" or "too_many_requests" or 2 more

            • "invalid_tool_input"

            • "unavailable"

            • "too_many_requests"

            • "execution_time_exceeded"

            • "output_file_too_large"

        • BetaBashCodeExecutionResultBlock object

          • type: "bash_code_execution_result"

            default: bash_code_execution_result

          • content: array of BetaBashCodeExecutionOutputBlock

            • type: "bash_code_execution_output"

              default: bash_code_execution_output

            • file_id: string

          • return_code: number

          • stderr: string

          • stdout: string

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

    • BetaTextEditorCodeExecutionToolResultBlock object

      • type: "text_editor_code_execution_tool_result"

        default: text_editor_code_execution_tool_result

      • content: BetaTextEditorCodeExecutionToolResultError or BetaTextEditorCodeExecutionViewResultBlock or BetaTextEditorCodeExecutionCreateResultBlock or BetaTextEditorCodeExecutionStrReplaceResultBlock

        • BetaTextEditorCodeExecutionToolResultError object

          • type: "text_editor_code_execution_tool_result_error"

            default: text_editor_code_execution_tool_result_error

          • error_code: "invalid_tool_input" or "unavailable" or "too_many_requests" or 2 more

            • "invalid_tool_input"

            • "unavailable"

            • "too_many_requests"

            • "execution_time_exceeded"

            • "file_not_found"

          • error_message: string or null

        • BetaTextEditorCodeExecutionViewResultBlock object

          • type: "text_editor_code_execution_view_result"

            default: text_editor_code_execution_view_result

          • content: string

          • file_type: "text" or "image" or "pdf"

            • "text"

            • "image"

            • "pdf"

          • num_lines: number or null

          • start_line: number or null

          • total_lines: number or null

        • BetaTextEditorCodeExecutionCreateResultBlock object

          • type: "text_editor_code_execution_create_result"

            default: text_editor_code_execution_create_result

          • is_file_update: boolean

        • BetaTextEditorCodeExecutionStrReplaceResultBlock object

          • type: "text_editor_code_execution_str_replace_result"

            default: text_editor_code_execution_str_replace_result

          • lines: array of string or null

          • new_lines: number or null

          • new_start: number or null

          • old_lines: number or null

          • old_start: number or null

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

    • BetaToolSearchToolResultBlock object

      • type: "tool_search_tool_result"

        default: tool_search_tool_result

      • content: BetaToolSearchToolResultError or BetaToolSearchToolSearchResultBlock

        • BetaToolSearchToolResultError object

          • type: "tool_search_tool_result_error"

            default: tool_search_tool_result_error

          • error_code: "invalid_tool_input" or "unavailable" or "too_many_requests" or "execution_time_exceeded"

            • "invalid_tool_input"

            • "unavailable"

            • "too_many_requests"

            • "execution_time_exceeded"

          • error_message: string or null

        • BetaToolSearchToolSearchResultBlock object

          • type: "tool_search_tool_search_result"

            default: tool_search_tool_search_result

          • tool_references: array of BetaToolReferenceBlock

            • type: "tool_reference"

              default: tool_reference

            • tool_name: string

              maxLength: 256, minLength: 1, pattern: ^[a-zA-Z0-9_-]{1,256}$

      • tool_use_id: string

        pattern: ^srvtoolu_[a-zA-Z0-9_]+$

    • BetaMCPToolUseBlock object

      • type: "mcp_tool_use"

        default: mcp_tool_use

      • id: string

        pattern: ^[a-zA-Z0-9_-]+$

      • input: map[unknown]

      • name: string

        The name of the MCP tool

      • server_name: string

        The name of the MCP server

    • BetaMCPToolResultBlock object

      • type: "mcp_tool_result"

        default: mcp_tool_result

      • content: string or array of BetaTextBlock

        • string

        • BetaMCPToolResultBlockContent = array of BetaTextBlock

          • type: "text"

            default: text

          • citations: array of BetaTextCitation or null

            Citations supporting the text block.

            The type of citation returned will depend on the type of document being cited. Citing a PDF results in page_location, plain text results in char_location, and content document results in content_block_location.

          • text: string

            minLength: 0

      • is_error: boolean

        default: false

      • tool_use_id: string

        pattern: ^[a-zA-Z0-9_-]+$

    • BetaContainerUploadBlock object

      Response model for a file uploaded to the container.

      • type: "container_upload"

        default: container_upload

      • file_id: string

    • BetaCompactionBlock object

      A compaction block returned when autocompact is triggered.

      When content is None, it indicates the compaction failed to produce a valid summary (e.g., malformed output from the model). Clients may round-trip compaction blocks with null content; the server treats them as no-ops.

      • type: "compaction"

        default: compaction

      • content: string or null

        Summary of compacted content, or null if compaction failed

      • encrypted_content: string or null

        Opaque metadata from prior compaction, to be round-tripped verbatim

    • BetaFallbackBlock object

      Marks the point in content where one model's output gives way to the next.

      One block appears per hop where a preceding model actually ran this turn and declined. A turn where no preceding model ran and declined has no such boundary and carries no block — the signal for whether a fallback model served the response is the presence of a fallback_message entry in usage.iterations, not this block.

      The block is treated like a server-tool content block for streaming: it arrives via the standard content_block_start / content_block_stop pair and carries no deltas.

      • type: "fallback"

        default: fallback

      • from: BetaFallbackInfo

        The model whose output ends at this point — the model that declined at this hop. When the declining hop is the requested model, its model echoes the top-level model string the caller sent (alias or canonical); when the declining hop is a fallback model, its model is that model's canonical id.

        • model: Model

          The model that will complete your prompt.

          See models for additional details and options.

          • "claude-fable-5-1" or "claude-mythos-5-1" or "claude-sonnet-5" or 14 more

            The model that will complete your prompt.

            See models for additional details and options.

            • "claude-fable-5-1"

              Frontier intelligence for ambitious tasks across coding, scientific discovery, and enterprise workflows

            • "claude-mythos-5-1"

              Our most capable model for cybersecurity and biology research, available through trusted access programs

            • "claude-sonnet-5"

              High-performance model for coding and agents

            • "claude-fable-5"

              Next generation of intelligence for the hardest knowledge work and coding problems

            • "claude-mythos-5"

              Most capable model for cybersecurity and biology research

            • "claude-opus-5"

              Powerful intelligence for long-running agents and coding

            • "claude-opus-4-8"

              Powerful intelligence for long-running agents and coding

            • "claude-opus-4-7"

              Powerful intelligence for long-running agents and coding

            • "claude-mythos-preview"

              New class of intelligence, strongest in coding and cybersecurity

            • "claude-opus-4-6"

              Powerful intelligence for long-running agents and coding

            • "claude-sonnet-4-6"

              Best combination of speed and intelligence

            • "claude-haiku-4-5"

              Fastest model with near-frontier intelligence

            • "claude-haiku-4-5-20251001"

              Fastest model with near-frontier intelligence

            • "claude-opus-4-5"

              Powerful intelligence for long-running agents and coding

            • "claude-opus-4-5-20251101"

              Powerful intelligence for long-running agents and coding

            • "claude-sonnet-4-5"

              High-performance model for agents and coding

            • "claude-sonnet-4-5-20250929"

              High-performance model for agents and coding

          • string

      • to: BetaFallbackInfo

        The fallback model producing the content that follows this block. Its model is always the canonical id.

      • trigger: BetaFallbackRefusalTrigger

        What caused the from model to hand over at this hop.

        • type: "refusal"

          default: refusal

        • category: "cyber" or "bio" or "frontier_llm" or 2 more or null

          The policy category that triggered a refusal.

          • "cyber"

            The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.

          • "bio"

            The request could enable biological harm, such as dangerous lab methods. Beneficial life sciences work can also trigger this category.

          • "frontier_llm"

            The request could assist the development of competing AI models, which is restricted under Anthropic's commercial terms. Benign machine learning work can also trigger this category.

          • "reasoning_extraction"

            The request asks the model to reproduce its internal reasoning in the response text. To get reasoning in a structured form instead, use adaptive thinking.

          • "general_harms"

            The request could be related to an area that was determined as harmful. Benign work might sometimes trigger this category.

  • context_management: BetaContextManagementResponse or null

    Context management response.

    Information about context management strategies applied during the request.

    • applied_edits: array of BetaClearToolUses20250919EditResponse or BetaClearThinking20251015EditResponse

      List of context management edits that were applied.

      • BetaClearToolUses20250919EditResponse object

        • type: "clear_tool_uses_20250919"

          The type of context management edit applied.

          default: clear_tool_uses_20250919

        • cleared_input_tokens: number

          Number of input tokens cleared by this edit.

          minimum: 0

        • cleared_tool_uses: number

          Number of tool uses that were cleared.

          minimum: 0

      • BetaClearThinking20251015EditResponse object

        • type: "clear_thinking_20251015"

          The type of context management edit applied.

          default: clear_thinking_20251015

        • cleared_input_tokens: number

          Number of input tokens cleared by this edit.

          minimum: 0

        • cleared_thinking_turns: number

          Number of thinking turns that were cleared.

          minimum: 0

  • diagnostics: BetaDiagnostics or null

    Response envelope for request-level diagnostics. Present (possibly null) whenever the caller supplied diagnostics on the request.

    • cache_miss_reason: BetaCacheMissModelChanged or BetaCacheMissSystemChanged or BetaCacheMissToolsChanged or 3 more or null

      Explains why the prompt cache could not fully reuse the prefix from the request identified by diagnostics.previous_message_id. null means diagnosis is still pending — the response was serialized before the background comparison completed.

      • BetaCacheMissModelChanged object

        • type: "model_changed"

          default: model_changed

        • cache_missed_input_tokens: number

          Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.

      • BetaCacheMissSystemChanged object

        • type: "system_changed"

          default: system_changed

        • cache_missed_input_tokens: number

          Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.

      • BetaCacheMissToolsChanged object

        • type: "tools_changed"

          default: tools_changed

        • cache_missed_input_tokens: number

          Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.

      • BetaCacheMissMessagesChanged object

        • type: "messages_changed"

          default: messages_changed

        • cache_missed_input_tokens: number

          Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.

      • BetaCacheMissPreviousMessageNotFound object

        • type: "previous_message_not_found"

          default: previous_message_not_found

      • BetaCacheMissUnavailable object

        • type: "unavailable"

          default: unavailable

  • model: Model

    The model that will complete your prompt.

    See models for additional details and options.

  • role: "assistant"

    Conversational role of the generated message.

    This will always be "assistant".

    default: assistant

  • stop_details: BetaRefusalStopDetails or null

    Structured information about a refusal.

    • type: "refusal"

      default: refusal

    • category: "cyber" or "bio" or "frontier_llm" or 2 more or null

      The policy category that triggered a refusal.

      • "cyber"

        The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.

      • "bio"

        The request could enable biological harm, such as dangerous lab methods. Beneficial life sciences work can also trigger this category.

      • "frontier_llm"

        The request could assist the development of competing AI models, which is restricted under Anthropic's commercial terms. Benign machine learning work can also trigger this category.

      • "reasoning_extraction"

        The request asks the model to reproduce its internal reasoning in the response text. To get reasoning in a structured form instead, use adaptive thinking.

      • "general_harms"

        The request could be related to an area that was determined as harmful. Benign work might sometimes trigger this category.

    • explanation: string or null

      Human-readable explanation of the refusal.

      This text is not guaranteed to be stable. null when no explanation is available for the category.

    • fallback_credit_token: string or null

      Opaque code that refunds the cache-miss cost when retrying this refused request on the fallback model. Pass it as fallback_credit_token on the retry request. Expires 5 minutes after the refusal.

      The retry is sent either with the same request body (system, messages, tools, and other render-shaping fields), or with the same body plus one appended assistant message whose content is the partial text (with any trailing whitespace stripped from the final text block) and paired server-tool blocks from this refusal — which also authorizes that appended turn as an assistant-prefill continuation on models that otherwise disallow prefill. A token minted mid-server-tool-loop whose partial content was continuable may only be redeemed the second way — if a same-body retry is rejected with a 400 saying the token must be redeemed by continuing the partial response, retry the second way instead. Either way: same workspace, same platform; a mismatch is a 400. Resending a token for an already-warm prefix is permitted but yields no additional credit.

      null when the refused model isn't eligible for a fallback credit.

    • fallback_has_prefill_claim: boolean or null

      Whether the accompanying fallback_credit_token may be redeemed with the appended-assistant retry form. Only set when fallback_credit_token is present.

      true: retry by resending the same request body plus one appended assistant message whose content is this response's content with any trailing whitespace stripped from the final text block and unpaired tool_use blocks omitted (the same appended-turn shape described on fallback_credit_token), with the token attached. false: retry by resending the original request body unchanged, with the token attached — the appended-assistant form is not available for this refusal (no continuable partial content, or the request uses output_format or a tool_choice that forces tool use). One exception: when the request used output_format or a forced tool_choice and the refusal arrived after server tools (including MCP connector tools) had already executed, the token may not be redeemable by either retry form; if the exact-body retry is then rejected with a 400 saying the token must be redeemed by continuing the partial response, discard the token and retry without it.

      Advisory: if an appended-assistant retry is rejected with a 400 despite true, fall back to resending the original request body with the token.

    • recommended_model: string or null

      The server's suggested retry target for this refusal. Populated when a fallback attempt could not be made (the fallback model's rate limit was exhausted, or it was overloaded); names the fallback model the caller can retry directly. Null otherwise.

  • stop_reason: BetaStopReason or null

    The reason that we stopped.

    This may be one the following values:

    • "end_turn": the model reached a natural stopping point
    • "max_tokens": we exceeded the requested max_tokens or the model's maximum
    • "stop_sequence": one of your provided custom stop_sequences was generated
    • "tool_use": the model invoked one or more tools
    • "pause_turn": we paused a long-running turn. You may provide the response back as-is in a subsequent request to let the model continue.
    • "refusal": when streaming classifiers intervene to handle potential policy violations
    • "model_context_window_exceeded": we exceeded the model's context window

    In non-streaming mode this value is always non-null. In streaming mode, it is null in the message_start event and non-null otherwise.

    • "end_turn"

    • "max_tokens"

    • "stop_sequence"

    • "tool_use"

    • "pause_turn"

    • "compaction"

    • "refusal"

    • "model_context_window_exceeded"

  • stop_sequence: string or null

    Which custom stop sequence was generated, if any.

    This value will be a non-null string if one of your custom stop sequences was generated.

  • usage: BetaUsage

    Billing and rate-limit usage.

    Anthropic's API bills and rate-limits by token counts, as tokens represent the underlying cost to our systems.

    Under the hood, the API transforms requests into a format suitable for the model. The model's output then goes through a parsing stage before becoming an API response. As a result, the token counts in usage will not match one-to-one with the exact visible content of an API request or response.

    For example, output_tokens will be non-zero, even for an empty string response from Claude.

    Total input tokens in a request is the summation of input_tokens, cache_creation_input_tokens, and cache_read_input_tokens.

    • cache_creation: BetaCacheCreation or null

      Breakdown of cached tokens by TTL

      • ephemeral_1h_input_tokens: number

        The number of input tokens used to create the 1 hour cache entry.

        default: 0, minimum: 0

      • ephemeral_5m_input_tokens: number

        The number of input tokens used to create the 5 minute cache entry.

        default: 0, minimum: 0

    • cache_creation_input_tokens: number or null

      The number of input tokens used to create the cache entry.

      minimum: 0

    • cache_read_input_tokens: number or null

      The number of input tokens read from the cache.

      minimum: 0

    • fallback_credit: BetaFallbackCreditUsage or null

      Outcome of the fallback_credit_token presented on this request.

      • status: BetaFallbackCreditRedeemed or BetaFallbackCreditNotApplied

        Whether the fallback-credit reprice was applied to this response's billing.

        A union discriminated on type. redeemed: the retry is billed as if the conversation had been on the retry model all along — including when the resulting shift is zero because there was nothing to move. not_applied: no reprice was applied; the arm's reason says why.

        • BetaFallbackCreditRedeemed object

          The reprice was applied: the retry is billed as if the conversation had been on the retry model all along.

          • type: "redeemed"

            default: redeemed

        • BetaFallbackCreditNotApplied object

          No reprice was applied; reason says why.

          • type: "not_applied"

            default: not_applied

          • reason: "body_mismatch" or "continuation_excluded" or "continuation_only" or 9 more

            Why the reprice was not applied.

            A closed enum; additions to the redemption-check vocabulary arrive as deliberate schema updates.

            • "body_mismatch"

            • "continuation_excluded"

            • "continuation_only"

            • "expired"

            • "invalid_target_model"

            • "not_enabled"

            • "reprice_unavailable"

            • "temporarily_unavailable"

            • "variant_fields_present"

            • "wrong_organization"

            • "wrong_platform"

            • "wrong_workspace"

          • remove_to_redeem: optional array of string or null

            Request fields to remove before retrying, so the retry can redeem this token.

            Present exactly when reason is variant_fields_present — never null, never an empty array; absent otherwise. Fields are named only from your own request, and only after the sealed variant hash matched. A served best-effort retry has already been billed at normal price; nothing redeems retroactively, but a corrected re-send inside the token's five-minute window can still redeem.

    • inference_geo: string or null

      The geographic region where inference was performed for this request.

    • input_tokens: number

      The number of input tokens which were used.

      minimum: 0

    • iterations: BetaIterationsUsage or null

      Per-iteration token usage breakdown.

      Each entry represents one sampling iteration, with its own input/output token counts and cache statistics, discriminated by type. For message entries (model sampling iterations, such as the turns of a server-side tool use loop), this allows you to:

      • Determine which iterations exceeded long context thresholds (>=200k tokens)
      • Calculate the context window size from the last message entry
      • Understand token accumulation across server-side tool use loops

      A compaction entry reports the token usage of the compaction operation itself — the server-side request that summarizes the context being closed — NOT the size of the context that was compacted away, and its token counts can be much smaller than that closed context (for example, a compaction that closes a ~200k-token context can report only a few thousand tokens). Do not derive the context window size from a compaction entry, even when it is the last entry. A compaction entry's tokens are not included in the top-level usage fields. When an input-token trigger is in effect (the default — 150,000 tokens unless configured otherwise), each compaction entry closes a context that had reached at least that threshold, though the context can exceed it by the final iteration's output and tool results.

      • BetaMessageIterationUsage object

        Token usage for a sampling iteration.

        • type: "message"

          Usage for a sampling iteration

          default: message

        • cache_creation: BetaCacheCreation or null

          Breakdown of cached tokens by TTL

        • cache_creation_input_tokens: number

          The number of input tokens used to create the cache entry.

          default: 0, minimum: 0

        • cache_read_input_tokens: number

          The number of input tokens read from the cache.

          default: 0, minimum: 0

        • input_tokens: number

          The number of input tokens which were used.

          minimum: 0

        • model: Model

          The model that will complete your prompt.

          See models for additional details and options.

        • output_tokens: number

          The number of output tokens which were used.

          minimum: 0

      • BetaCompactionIterationUsage object

        Token usage for a compaction iteration.

        • type: "compaction"

          Usage for a compaction iteration

          default: compaction

        • cache_creation: BetaCacheCreation or null

          Breakdown of cached tokens by TTL

        • cache_creation_input_tokens: number

          The number of input tokens used to create the cache entry.

          default: 0, minimum: 0

        • cache_read_input_tokens: number

          The number of input tokens read from the cache.

          default: 0, minimum: 0

        • input_tokens: number

          The number of input tokens which were used.

          minimum: 0

        • output_tokens: number

          The number of output tokens which were used.

          minimum: 0

      • BetaAdvisorMessageIterationUsage object

        Token usage for an advisor sub-inference iteration.

        • type: "advisor_message"

          Usage for an advisor sub-inference iteration

          default: advisor_message

        • cache_creation: BetaCacheCreation or null

          Breakdown of cached tokens by TTL

        • cache_creation_input_tokens: number

          The number of input tokens used to create the cache entry.

          default: 0, minimum: 0

        • cache_read_input_tokens: number

          The number of input tokens read from the cache.

          default: 0, minimum: 0

        • input_tokens: number

          The number of input tokens which were used.

          minimum: 0

        • model: Model

          The model that will complete your prompt.

          See models for additional details and options.

        • output_tokens: number

          The number of output tokens which were used.

          minimum: 0

      • BetaFallbackMessageIterationUsage object

        Token usage for the fallback-model attempt of a server-side fallback request.

        Produced in place of a message entry for whichever hop served the response. A declined hop produces the existing message entry. Whether a fallback model served the response is signalled by the presence of this entry in usage.iterations.

        • type: "fallback_message"

          Usage for the fallback-model attempt that served the response

          default: fallback_message

        • cache_creation: BetaCacheCreation or null

          Breakdown of cached tokens by TTL

        • cache_creation_input_tokens: number

          The number of input tokens used to create the cache entry.

          default: 0, minimum: 0

        • cache_read_input_tokens: number

          The number of input tokens read from the cache.

          default: 0, minimum: 0

        • input_tokens: number

          The number of input tokens which were used.

          minimum: 0

        • model: Model

          The model that will complete your prompt.

          See models for additional details and options.

        • output_tokens: number

          The number of output tokens which were used.

          minimum: 0

    • output_tokens: number

      The number of output tokens which were used.

      minimum: 0

    • output_tokens_details: BetaOutputTokensDetails or null

      Breakdown of output tokens by category.

      output_tokens remains the inclusive, authoritative total used for billing. This object provides a read-only decomposition for observability — for example, how many of the billed output tokens were spent on internal reasoning that may have been summarized before being returned to you.

      • thinking_tokens: number

        Number of output tokens the model generated as internal reasoning, including the thinking-block delimiter tokens.

        Reflects the raw reasoning the model produced, not the (possibly shorter) summarized thinking text returned in the response body. Computed by re-tokenizing the raw reasoning text, so it may differ from the model's exact generation count by a small number of tokens. Always ≤ output_tokens; output_tokens - thinking_tokens approximates the non-reasoning output.

        default: 0, minimum: 0

    • server_tool_use: BetaServerToolUsage or null

      The number of server tool requests.

      • web_fetch_requests: number

        The number of web fetch tool requests.

        default: 0, minimum: 0

      • web_search_requests: number

        The number of web search tool requests.

        default: 0, minimum: 0

    • service_tier: "standard" or "priority" or "batch" or null

      If the request used the priority, standard, or batch tier.

      • "standard"

      • "priority"

      • "batch"

    • speed: "standard" or "fast" or null

      Inference speed mode. fast provides significantly faster output token generation at premium pricing. Not all models support fast; invalid combinations are rejected at create time.

      • "standard"

      • "fast"

  • input_transformations: optional array of BetaThinkingDroppedInputTransformation or null

    Changes the API made to the request's input before showing it to the model: one entry per change, in request order. Today the only entry type is thinking_dropped — a thinking, redacted_thinking or connector_text block from the request's messages that was removed from the prompt instead of being shown to the model because it failed a binding check. More entry types may be added over time; ignore types you do not recognize.

    Requires anthropic-beta: thinking-binding-controls-2026-08-01. Present on every such response from a model that supports extended thinking, as [] when nothing was changed; without the beta, blocks are removed all the same but nothing is reported. Removed blocks contribute nothing to usage.input_tokens. When streaming, the array is final in message_start; the final message_delta event carries it only when a server-side model fallback happened mid-stream, in which case it holds the serving model's entries and replaces the one in message_start.

    • type: "thinking_dropped"

      Always thinking_dropped for this entry type.

      default: thinking_dropped

    • path: string

      Where the removed block was in your request, as messages.\{i\}.content.\{j\}: i indexes the messages array you sent and j that message's content array — the same form error messages use.

    • reason: "model_binding_mismatch" or "prefix_binding_mismatch" or "organization_binding_mismatch" or "end_user_binding_mismatch"

      Which binding check removed the block: model_binding_mismatch — it was created by a model whose reasoning the requested model may not read; prefix_binding_mismatch — the conversation before it differs from the conversation it was created in (the rest of that turn's consecutive thinking blocks are removed with it, each with this reason); organization_binding_mismatch — it was created under a different organization (an Anthropic organization, AWS account or Google Cloud project) and this organization is not one of its additional organizations; end_user_binding_mismatch — it was created for a different end user, or was removed by the consumer-organization binding. A block that would fail several checks reports one reason, in this order of precedence: organization_binding_mismatch, end_user_binding_mismatch, model_binding_mismatch, prefix_binding_mismatch.

      • "model_binding_mismatch"

      • "prefix_binding_mismatch"

      • "organization_binding_mismatch"

      • "end_user_binding_mismatch"

• BetaRawMessageStreamEvent = BetaRawMessageStartEvent or BetaRawMessageDeltaEvent or BetaRawMessageStopEvent or 3 more

  • BetaRawMessageStartEvent object

    • type: "message_start"

      default: message_start

    • message: BetaMessage

  • BetaRawMessageDeltaEvent object

    • type: "message_delta"

      default: message_delta

    • context_management: BetaContextManagementResponse or null

      Information about context management strategies applied during the request

    • delta: object

      • container: BetaContainer or null

        Information about the container used in the request (for the code execution tool)

      • stop_details: BetaRefusalStopDetails or null

        Structured information about a refusal.

      • stop_reason: BetaStopReason or null

      • stop_sequence: string or null

    • usage: BetaMessageDeltaUsage

      Billing and rate-limit usage.

      Anthropic's API bills and rate-limits by token counts, as tokens represent the underlying cost to our systems.

      Under the hood, the API transforms requests into a format suitable for the model. The model's output then goes through a parsing stage before becoming an API response. As a result, the token counts in usage will not match one-to-one with the exact visible content of an API request or response.

      For example, output_tokens will be non-zero, even for an empty string response from Claude.

      Total input tokens in a request is the summation of input_tokens, cache_creation_input_tokens, and cache_read_input_tokens.

      • cache_creation_input_tokens: number or null

        The cumulative number of input tokens used to create the cache entry.

        minimum: 0

      • cache_read_input_tokens: number or null

        The cumulative number of input tokens read from the cache.

        minimum: 0

      • fallback_credit: BetaFallbackCreditUsage or null

        Outcome of the fallback_credit_token presented on this request.

      • input_tokens: number or null

        The cumulative number of input tokens which were used.

        minimum: 0

      • iterations: BetaIterationsUsage or null

        Per-iteration token usage breakdown.

        Each entry represents one sampling iteration, with its own input/output token counts and cache statistics, discriminated by type. For message entries (model sampling iterations, such as the turns of a server-side tool use loop), this allows you to:

        • Determine which iterations exceeded long context thresholds (>=200k tokens)
        • Calculate the context window size from the last message entry
        • Understand token accumulation across server-side tool use loops

        A compaction entry reports the token usage of the compaction operation itself — the server-side request that summarizes the context being closed — NOT the size of the context that was compacted away, and its token counts can be much smaller than that closed context (for example, a compaction that closes a ~200k-token context can report only a few thousand tokens). Do not derive the context window size from a compaction entry, even when it is the last entry. A compaction entry's tokens are not included in the top-level usage fields. When an input-token trigger is in effect (the default — 150,000 tokens unless configured otherwise), each compaction entry closes a context that had reached at least that threshold, though the context can exceed it by the final iteration's output and tool results.

      • output_tokens: number

        The cumulative number of output tokens which were used.

      • output_tokens_details: BetaOutputTokensDetails or null

        Breakdown of output tokens by category.

        output_tokens remains the inclusive, authoritative total used for billing. This object provides a read-only decomposition for observability — for example, how many of the billed output tokens were spent on internal reasoning that may have been summarized before being returned to you.

      • server_tool_use: BetaServerToolUsage or null

        The number of server tool requests.

    • input_transformations: optional array of BetaThinkingDroppedInputTransformation or null

      Changes the API made to the request's input before showing it to the model: one entry per change, in request order. Today the only entry type is thinking_dropped — a thinking, redacted_thinking or connector_text block from the request's messages that was removed from the prompt instead of being shown to the model because it failed a binding check. More entry types may be added over time; ignore types you do not recognize.

      Requires anthropic-beta: thinking-binding-controls-2026-08-01. Present on every such response from a model that supports extended thinking, as [] when nothing was changed; without the beta, blocks are removed all the same but nothing is reported. Removed blocks contribute nothing to usage.input_tokens. When streaming, the array is final in message_start; the final message_delta event carries it only when a server-side model fallback happened mid-stream, in which case it holds the serving model's entries and replaces the one in message_start.

      • type: "thinking_dropped"

        Always thinking_dropped for this entry type.

        default: thinking_dropped

      • path: string

        Where the removed block was in your request, as messages.\{i\}.content.\{j\}: i indexes the messages array you sent and j that message's content array — the same form error messages use.

      • reason: "model_binding_mismatch" or "prefix_binding_mismatch" or "organization_binding_mismatch" or "end_user_binding_mismatch"

        Which binding check removed the block: model_binding_mismatch — it was created by a model whose reasoning the requested model may not read; prefix_binding_mismatch — the conversation before it differs from the conversation it was created in (the rest of that turn's consecutive thinking blocks are removed with it, each with this reason); organization_binding_mismatch — it was created under a different organization (an Anthropic organization, AWS account or Google Cloud project) and this organization is not one of its additional organizations; end_user_binding_mismatch — it was created for a different end user, or was removed by the consumer-organization binding. A block that would fail several checks reports one reason, in this order of precedence: organization_binding_mismatch, end_user_binding_mismatch, model_binding_mismatch, prefix_binding_mismatch.

  • BetaRawMessageStopEvent object

    • type: "message_stop"

      default: message_stop

  • BetaRawContentBlockStartEvent object

    • type: "content_block_start"

      default: content_block_start

    • content_block: BetaTextBlock or BetaThinkingBlock or BetaRedactedThinkingBlock or 14 more

      Response model for a file uploaded to the container.

      • BetaTextBlock object

      • BetaThinkingBlock object

      • BetaRedactedThinkingBlock object

      • BetaToolUseBlock object

      • BetaServerToolUseBlock object

      • BetaWebSearchToolResultBlock object

      • BetaWebFetchToolResultBlock object

      • BetaAdvisorToolResultBlock object

      • BetaCodeExecutionToolResultBlock object

      • BetaBashCodeExecutionToolResultBlock object

      • BetaTextEditorCodeExecutionToolResultBlock object

      • BetaToolSearchToolResultBlock object

      • BetaMCPToolUseBlock object

      • BetaMCPToolResultBlock object

      • BetaContainerUploadBlock object

        Response model for a file uploaded to the container.

      • BetaCompactionBlock object

        A compaction block returned when autocompact is triggered.

        When content is None, it indicates the compaction failed to produce a valid summary (e.g., malformed output from the model). Clients may round-trip compaction blocks with null content; the server treats them as no-ops.

      • BetaFallbackBlock object

        Marks the point in content where one model's output gives way to the next.

        One block appears per hop where a preceding model actually ran this turn and declined. A turn where no preceding model ran and declined has no such boundary and carries no block — the signal for whether a fallback model served the response is the presence of a fallback_message entry in usage.iterations, not this block.

        The block is treated like a server-tool content block for streaming: it arrives via the standard content_block_start / content_block_stop pair and carries no deltas.

    • index: number

  • BetaRawContentBlockDeltaEvent object

    • type: "content_block_delta"

      default: content_block_delta

    • delta: BetaRawContentBlockDelta

      • BetaTextDelta object

        • type: "text_delta"

          default: text_delta

        • text: string

      • BetaInputJSONDelta object

        • type: "input_json_delta"

          default: input_json_delta

        • partial_json: string

      • BetaCitationsDelta object

        • type: "citations_delta"

          default: citations_delta

        • citation: BetaCitationCharLocation or BetaCitationPageLocation or BetaCitationContentBlockLocation or 2 more

          • BetaCitationCharLocation object

          • BetaCitationPageLocation object

          • BetaCitationContentBlockLocation object

          • BetaCitationsWebSearchResultLocation object

          • BetaCitationSearchResultLocation object

      • BetaThinkingDelta object

        • type: "thinking_delta"

          default: thinking_delta

        • estimated_tokens: number or null

          Per-frame increment of a coarse, running estimate of the tokens this thinking block has produced so far. Present whenever the thinking-token-count-2026-05-13 beta is set; null unless thinking.display resolves to "omitted" and a count is due this frame. Sum the increments across thinking_delta frames on this block for a progress indicator. Each increment is a non-negative multiple of a fixed quantum and the cadence is rate-limited, so this is a deliberately lossy display hint, not a billable count; usage.output_tokens remains authoritative.

        • thinking: string

          The incremental thinking text for this content block. Concatenate the thinking values of successive thinking_delta events to assemble the block's full thinking value.

      • BetaSignatureDelta object

        • type: "signature_delta"

          default: signature_delta

        • signature: string

          The signature for this thinking block: an opaque value used to verify that the block was generated by Claude when it is passed back to the API. Delivered in a signature_delta event just before the block's content_block_stop event.

      • BetaCompactionContentBlockDelta object

        • type: "compaction_delta"

          default: compaction_delta

        • content: string or null

        • encrypted_content: string or null

          Opaque metadata from prior compaction, to be round-tripped verbatim

    • index: number

  • BetaRawContentBlockStopEvent object

    • type: "content_block_stop"

      default: content_block_stop

    • index: number