mozex/anthropic-php

What's New

This release brings the response layer up to date with the Anthropic API changes shipped between April and July 2026: new usage fields, server-side fallback, compaction, and cache diagnostics.

New usage fields

Three additions to usage on Messages responses:

$response->usage->outputTokensDetails?->thinkingTokens; // how many billed output tokens were reasoning
$response->usage->speed;                                // 'fast' or 'standard' on fast mode requests
$response->usage->iterations;                           // per-attempt records under the fallback and compaction betas

When streaming, outputTokensDetails arrives on the final message_delta event.

Server-side fallback (beta)

When a refused request is retried on another model (the fallbacks parameter with the server-side-fallback-2026-07-01 beta header), the response now parses fully:

• fallback content blocks mark the handoff: $block->from->model and $block->to->model
• usage->iterations records every attempt: a message entry for a declined model, a fallback_message entry for the model that served the turn
• stop_details->recommended_model names a model to retry directly when the fallback itself was rate limited
• stop_details->category now covers frontier_llm, reasoning_extraction, and general_harms alongside cyber and bio

Compaction (beta)

Under the compact-2026-01-12 beta, responses can include a compaction content block whose content is a string summary. The SDK now handles that block, the compaction_delta stream event, the top-level context_management response field, and the stop_reason: 'compaction' value that appears with pause_after_compaction. All of it was verified against live API responses, including the quirk that iteration entries under this beta carry no model key.

foreach ($response->content as $block) {
    if ($block->type === 'compaction') {
        $block->content; // 'Summary of the conversation: ...'
    }
}

$response->context_management; // ['applied_edits' => [...]] or null

Cache diagnostics (beta)

With the cache-diagnosis-2026-04-07 header and a diagnostics request parameter, the API reports where two consecutive requests diverged:

$response->diagnostics?->cache_miss_reason?->type;                      // 'system_changed'
$response->diagnostics?->cache_miss_reason?->cache_missed_input_tokens; // 41850

On streams, diagnostics arrives on the message_start event.

Documentation

The Messages, Thinking, Streaming, and Server Tools guides now cover refusal handling with fallback, compaction, cache diagnostics, thinking token usage, and the latest server tool versions (web_search_20260318, web_fetch_20260318, code_execution_20260521).

Full Changelog: 1.7.0...1.8.0

What's New

This release brings the response layer up to date with the Anthropic API changes shipped between April and July 2026: new usage fields, server-side fallback, compaction, and cache diagnostics.

New usage fields

Three additions to usage on Messages responses:

$response->usage->outputTokensDetails?->thinkingTokens; // how many billed output tokens were reasoning
$response->usage->speed;                                // 'fast' or 'standard' on fast mode requests
$response->usage->iterations;                           // per-attempt records under the fallback and compaction betas

When streaming, outputTokensDetails arrives on the final message_delta event.

Server-side fallback (beta)

When a refused request is retried on another model (the fallbacks parameter with the server-side-fallback-2026-07-01 beta header), the response now parses fully:

• fallback content blocks mark the handoff: $block->from->model and $block->to->model
• usage->iterations records every attempt: a message entry for a declined model, a fallback_message entry for the model that served the turn
• stop_details->recommended_model names a model to retry directly when the fallback itself was rate limited
• stop_details->category now covers frontier_llm, reasoning_extraction, and general_harms alongside cyber and bio

Compaction (beta)

Under the compact-2026-01-12 beta, responses can include a compaction content block whose content is a string summary. The SDK now handles that block, the compaction_delta stream event, the top-level context_management response field, and the stop_reason: 'compaction' value that appears with pause_after_compaction. All of it was verified against live API responses, including the quirk that iteration entries under this beta carry no model key.

foreach ($response->content as $block) {
    if ($block->type === 'compaction') {
        $block->content; // 'Summary of the conversation: ...'
    }
}

$response->context_management; // ['applied_edits' => [...]] or null

Cache diagnostics (beta)

With the cache-diagnosis-2026-04-07 header and a diagnostics request parameter, the API reports where two consecutive requests diverged:

$response->diagnostics?->cache_miss_reason?->type;                      // 'system_changed'
$response->diagnostics?->cache_miss_reason?->cache_missed_input_tokens; // 41850

On streams, diagnostics arrives on the message_start event.

Documentation

The Messages, Thinking, Streaming, and Server Tools guides now cover refusal handling with fallback, compaction, cache diagnostics, thinking token usage, and the latest server tool versions (web_search_20260318, web_fetch_20260318, code_execution_20260521).

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.7.0...1.8.0

What's New

This release brings the response layer up to date with the Anthropic API changes shipped between April and July 2026: new usage fields, server-side fallback, compaction, and cache diagnostics.

New usage fields

Three additions to usage on Messages responses:

$response -> usage -> outputTokensDetails ?-> thinkingTokens; // how many billed output tokens were reasoning $response -> usage -> speed; // 'fast' or 'standard' on fast mode requests $response -> usage -> iterations; // per-attempt records under the fallback and compaction betas

When streaming, outputTokensDetails arrives on the final message_delta event.

Server-side fallback (beta)

When a refused request is retried on another model (the fallbacks parameter with the server-side-fallback-2026-07-01 beta header), the response now parses fully:

• fallback content blocks mark the handoff: $block->from->model and $block->to->model

• usage->iterations records every attempt: a message entry for a declined model, a fallback_message entry for the model that served the turn

• stop_details->recommended_model names a model to retry directly when the fallback itself was rate limited

• stop_details->category now covers frontier_llm , reasoning_extraction , and general_harms alongside cyber and bio

Compaction (beta)

Under the compact-2026-01-12 beta, responses can include a compaction content block whose content is a string summary. The SDK now handles that block, the compaction_delta stream event, the top-level context_management response field, and the stop_reason: 'compaction' value that appears with pause_after_compaction . All of it was verified against live API responses, including the quirk that iteration entries under this beta carry no model key.

foreach ($response -> content as $block) { if ($block -> type === 'compaction' ) { $block -> content; // 'Summary of the conversation: ...' } } $response -> context_management; // ['applied_edits' => [...]] or null

Cache diagnostics (beta)

With the cache-diagnosis-2026-04-07 header and a diagnostics request parameter, the API reports where two consecutive requests diverged:

$response -> diagnostics ?-> cache_miss_reason ?-> type; // 'system_changed' $response -> diagnostics ?-> cache_miss_reason ?-> cache_missed_input_tokens; // 41850

On streams, diagnostics arrives on the message_start event.

Documentation

The Messages , Thinking , Streaming , and Server Tools guides now cover refusal handling with fallback, compaction, cache diagnostics, thinking token usage, and the latest server tool versions ( web_search_20260318 , web_fetch_20260318 , code_execution_20260521 ).

Full Changelog : 1.7.0...1.8.0

What's New

Files API

Upload a document once and reference it by file_id on later Messages calls. Useful for repeated PDFs and images, and for reading outputs produced by the code execution tool and Skills.

Five methods on $client->files():

• upload(array $parameters): multipart upload, returns a FileResponse
• list(array $parameters = []): cursor-paginated listing
• retrieveMetadata(string $fileId): fetch metadata for a single file
• download(string $fileId): raw bytes for files produced by code execution or Skills (user-uploaded files are not downloadable)
• delete(string $fileId): returns a DeletedFileResponse

$file = $client->files()->upload([
    'file' => fopen('/path/to/doc.pdf', 'r'),
]);

$response = $client->messages()->create([
    'model' => 'claude-opus-4-6',
    'max_tokens' => 1024,
    'betas' => ['files-api-2025-04-14'],
    'messages' => [[
        'role' => 'user',
        'content' => [
            ['type' => 'text', 'text' => 'Summarise this.'],
            ['type' => 'document', 'source' => ['type' => 'file', 'file_id' => $file->id]],
        ],
    ]],
]);

Anthropic currently flags the Files endpoints as beta. The SDK auto-injects the required anthropic-beta: files-api-2025-04-14 header on every $client->files() call, so you don't type the version string. When you reference a file_id inside a Messages call, pass 'betas' => ['files-api-2025-04-14'] on that call too; the Messages endpoint also needs the header when a file is referenced.

Documentation

New Files usage guide covers upload, list, retrieve, download, delete, and Messages integration, including the per-call betas pattern for referencing uploaded files.

Testing

Every Files response DTO has a fake() method for use with ClientFake:

• FileResponse::fake()
• FileListResponse::fake()
• DeletedFileResponse::fake()

What's Changed

• Add Files API resource by @mozex in #17

Full Changelog: 1.6.0...1.7.0

What's New

Files API

Upload a document once and reference it by file_id on later Messages calls. Useful for repeated PDFs and images, and for reading outputs produced by the code execution tool and Skills.

Five methods on $client->files():

• upload(array $parameters): multipart upload, returns a FileResponse
• list(array $parameters = []): cursor-paginated listing
• retrieveMetadata(string $fileId): fetch metadata for a single file
• download(string $fileId): raw bytes for files produced by code execution or Skills (user-uploaded files are not downloadable)
• delete(string $fileId): returns a DeletedFileResponse

$file = $client->files()->upload([
    'file' => fopen('/path/to/doc.pdf', 'r'),
]);

$response = $client->messages()->create([
    'model' => 'claude-opus-4-6',
    'max_tokens' => 1024,
    'betas' => ['files-api-2025-04-14'],
    'messages' => [[
        'role' => 'user',
        'content' => [
            ['type' => 'text', 'text' => 'Summarise this.'],
            ['type' => 'document', 'source' => ['type' => 'file', 'file_id' => $file->id]],
        ],
    ]],
]);

Anthropic currently flags the Files endpoints as beta. The SDK auto-injects the required anthropic-beta: files-api-2025-04-14 header on every $client->files() call, so you don't type the version string. When you reference a file_id inside a Messages call, pass 'betas' => ['files-api-2025-04-14'] on that call too; the Messages endpoint also needs the header when a file is referenced.

Documentation

New Files usage guide covers upload, list, retrieve, download, delete, and Messages integration, including the per-call betas pattern for referencing uploaded files.

Testing

Every Files response DTO has a fake() method for use with ClientFake:

• FileResponse::fake()
• FileListResponse::fake()
• DeletedFileResponse::fake()

What's Changed

• Add Files API resource by @mozex in https://github.com/mozex/anthropic-php/pull/17

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.6.0...1.7.0

What's New

Files API

Upload a document once and reference it by file_id on later Messages calls. Useful for repeated PDFs and images, and for reading outputs produced by the code execution tool and Skills.

Five methods on $client->files() :

• upload(array $parameters) : multipart upload, returns a FileResponse

• list(array $parameters = []) : cursor-paginated listing

• retrieveMetadata(string $fileId) : fetch metadata for a single file

• download(string $fileId) : raw bytes for files produced by code execution or Skills (user-uploaded files are not downloadable)

• delete(string $fileId) : returns a DeletedFileResponse

$file = $client -> files () -> upload ([ 'file' => fopen ( '/path/to/doc.pdf' , 'r' ), ]); $response = $client -> messages () -> create ([ 'model' => 'claude-opus-4-6' , 'max_tokens' => 1024 , 'betas' => [ 'files-api-2025-04-14' ], 'messages' => [[ 'role' => 'user' , 'content' => [ [ 'type' => 'text' , 'text' => 'Summarise this.' ], [ 'type' => 'document' , 'source' => [ 'type' => 'file' , 'file_id' => $file -> id]], ], ]], ]);

Anthropic currently flags the Files endpoints as beta. The SDK auto-injects the required anthropic-beta: files-api-2025-04-14 header on every $client->files() call, so you don't type the version string. When you reference a file_id inside a Messages call, pass 'betas' => ['files-api-2025-04-14'] on that call too; the Messages endpoint also needs the header when a file is referenced.

Documentation

New Files usage guide covers upload, list, retrieve, download, delete, and Messages integration, including the per-call betas pattern for referencing uploaded files.

Testing

Every Files response DTO has a fake() method for use with ClientFake :

• FileResponse::fake()

• FileListResponse::fake()

• DeletedFileResponse::fake()

What's Changed

• Add Files API resource by @mozex in #17

Full Changelog : 1.6.0...1.7.0

What's Changed

• Add betas parameter for per-request beta feature opt-in by @mozex in #16

Full Changelog: 1.5.0...1.6.0

What's Changed

• Add betas parameter for per-request beta feature opt-in by @mozex in https://github.com/mozex/anthropic-php/pull/16

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.5.0...1.6.0

What's Changed

• Add betas parameter for per-request beta feature opt-in by @mozex in #16

Full Changelog : 1.5.0...1.6.0

What's Changed

Added

Models API

• Add maxInputTokens and maxTokens properties to RetrieveResponse (and every item returned by ListResponse)
• Add typed capabilities tree on model responses: batch, citations, codeExecution, imageInput, pdfInput, structuredOutputs, thinking (with types.adaptive and types.enabled), and effort (with low, medium, high, max levels)
• Add capabilities.contextManagement as a map keyed by the raw API strategy name (clear_thinking_20251015, clear_tool_uses_20250919, compact_20260112, and any future version), so new strategies Anthropic ships are captured automatically without a package update

Messages API

• Add stop_details on CreateResponse with type, category, and explanation — populated when stop_reason is 'refusal'
• Add caller on tool_use and server_tool_use content blocks (type and tool_id) so direct model calls can be distinguished from calls made inside a code execution sandbox
• Add container_upload content block support with file_id
• Add inferenceGeo to CreateResponseUsage and CreateStreamedResponseUsage, surfacing which region handled the request ('us', 'eu', or null)
• Document the new pause_turn and refusal stop reasons and the idiom for resuming paused turns

Meta information

• Add priorityInputTokenLimit and priorityOutputTokenLimit on MetaInformation, parsing the six anthropic-priority-* headers into typed properties instead of dropping them into the generic custom bucket

Documentation

• Ship the full documentation site at mozex.dev/docs/anthropic-php/v1 with dedicated pages for every feature: introduction, configuration, messages, tool use, server tools, streaming, thinking, citations, batches, token counting, models, completions, meta information, testing, and error handling
• Every page now links back to the matching Anthropic reference on platform.claude.com/docs/

Improved

• tool_use / server_tool_use toArray() output now includes the caller object when present, so response round-trips stay lossless
• MetaInformation rate-limit headers (including the new priority tier) round-trip cleanly through toArray()

Full Changelog: 1.4.0...1.5.0

What's Changed

Added

Models API

• Add maxInputTokens and maxTokens properties to RetrieveResponse (and every item returned by ListResponse)
• Add typed capabilities tree on model responses: batch, citations, codeExecution, imageInput, pdfInput, structuredOutputs, thinking (with types.adaptive and types.enabled), and effort (with low, medium, high, max levels)
• Add capabilities.contextManagement as a map keyed by the raw API strategy name (clear_thinking_20251015, clear_tool_uses_20250919, compact_20260112, and any future version), so new strategies Anthropic ships are captured automatically without a package update

Messages API

• Add stop_details on CreateResponse with type, category, and explanation — populated when stop_reason is 'refusal'
• Add caller on tool_use and server_tool_use content blocks (type and tool_id) so direct model calls can be distinguished from calls made inside a code execution sandbox
• Add container_upload content block support with file_id
• Add inferenceGeo to CreateResponseUsage and CreateStreamedResponseUsage, surfacing which region handled the request ('us', 'eu', or null)
• Document the new pause_turn and refusal stop reasons and the idiom for resuming paused turns

Meta information

• Add priorityInputTokenLimit and priorityOutputTokenLimit on MetaInformation, parsing the six anthropic-priority-* headers into typed properties instead of dropping them into the generic custom bucket

Documentation

• Ship the full documentation site at mozex.dev/docs/anthropic-php/v1 with dedicated pages for every feature: introduction, configuration, messages, tool use, server tools, streaming, thinking, citations, batches, token counting, models, completions, meta information, testing, and error handling
• Every page now links back to the matching Anthropic reference on platform.claude.com/docs/

Improved

• tool_use / server_tool_use toArray() output now includes the caller object when present, so response round-trips stay lossless
• MetaInformation rate-limit headers (including the new priority tier) round-trip cleanly through toArray()

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.4.0...1.5.0

What's Changed

Added

Models API

• Add maxInputTokens and maxTokens properties to RetrieveResponse (and every item returned by ListResponse )

• Add typed capabilities tree on model responses: batch , citations , codeExecution , imageInput , pdfInput , structuredOutputs , thinking (with types.adaptive and types.enabled ), and effort (with low , medium , high , max levels)

• Add capabilities.contextManagement as a map keyed by the raw API strategy name ( clear_thinking_20251015 , clear_tool_uses_20250919 , compact_20260112 , and any future version), so new strategies Anthropic ships are captured automatically without a package update

Messages API

• Add stop_details on CreateResponse with type , category , and explanation — populated when stop_reason is 'refusal'

• Add caller on tool_use and server_tool_use content blocks ( type and tool_id ) so direct model calls can be distinguished from calls made inside a code execution sandbox

• Add container_upload content block support with file_id

• Add inferenceGeo to CreateResponseUsage and CreateStreamedResponseUsage , surfacing which region handled the request ( 'us' , 'eu' , or null )

• Document the new pause_turn and refusal stop reasons and the idiom for resuming paused turns

Meta information

• Add priorityInputTokenLimit and priorityOutputTokenLimit on MetaInformation , parsing the six anthropic-priority-* headers into typed properties instead of dropping them into the generic custom bucket

Documentation

• Ship the full documentation site at mozex.dev/docs/anthropic-php/v1 with dedicated pages for every feature: introduction, configuration, messages, tool use, server tools, streaming, thinking, citations, batches, token counting, models, completions, meta information, testing, and error handling

• Every page now links back to the matching Anthropic reference on platform.claude.com/docs/

Improved

• tool_use / server_tool_use toArray() output now includes the caller object when present, so response round-trips stay lossless

• MetaInformation rate-limit headers (including the new priority tier) round-trip cleanly through toArray()

Full Changelog : 1.4.0...1.5.0

What's Changed

Added

• Add adaptive thinking support (thinking.type: "adaptive") for Claude Opus 4.6 and Sonnet 4.6
• Add display option on thinking config (summarized / omitted) to control thinking output in responses
• Add output_config.effort parameter for guiding thinking depth (max, high, medium, low)
• Add server_tool_use content block type for server-side tools (web search, web fetch, code execution, tool search)
• Add web_search_tool_result content block type with search results array
• Add web_fetch_tool_result content block type
• Add code_execution_tool_result, bash_code_execution_tool_result, and text_editor_code_execution_tool_result content block types
• Add tool_search_tool_result content block type
• Add tool_use_id and content properties on CreateResponseContent for all server tool result blocks
• Add container field on CreateResponse for code execution sandbox persistence
• Add citations array on text content blocks, supporting all 5 citation location types: char_location, page_location, content_block_location, web_search_result_location, search_result_location
• Add citations_delta streaming support with citation property on CreateStreamedResponseDelta
• Add webFetchRequests, codeExecutionRequests, and toolSearchRequests to CreateResponseUsageServerToolUse
• Add streaming support for server_tool_use and server tool result blocks in CreateStreamedResponseContentBlockStart

Full Changelog: 1.3.3...1.4.0

What's Changed

Added

• Add adaptive thinking support (thinking.type: "adaptive") for Claude Opus 4.6 and Sonnet 4.6
• Add display option on thinking config (summarized / omitted) to control thinking output in responses
• Add output_config.effort parameter for guiding thinking depth (max, high, medium, low)
• Add server_tool_use content block type for server-side tools (web search, web fetch, code execution, tool search)
• Add web_search_tool_result content block type with search results array
• Add web_fetch_tool_result content block type
• Add code_execution_tool_result, bash_code_execution_tool_result, and text_editor_code_execution_tool_result content block types
• Add tool_search_tool_result content block type
• Add tool_use_id and content properties on CreateResponseContent for all server tool result blocks
• Add container field on CreateResponse for code execution sandbox persistence
• Add citations array on text content blocks, supporting all 5 citation location types: char_location, page_location, content_block_location, web_search_result_location, search_result_location
• Add citations_delta streaming support with citation property on CreateStreamedResponseDelta
• Add webFetchRequests, codeExecutionRequests, and toolSearchRequests to CreateResponseUsageServerToolUse
• Add streaming support for server_tool_use and server tool result blocks in CreateStreamedResponseContentBlockStart

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.3.3...1.4.0

What's Changed

Added

• Add adaptive thinking support ( thinking.type: "adaptive" ) for Claude Opus 4.6 and Sonnet 4.6

• Add display option on thinking config ( summarized / omitted ) to control thinking output in responses

• Add output_config.effort parameter for guiding thinking depth ( max , high , medium , low )

• Add server_tool_use content block type for server-side tools (web search, web fetch, code execution, tool search)

• Add web_search_tool_result content block type with search results array

• Add web_fetch_tool_result content block type

• Add code_execution_tool_result , bash_code_execution_tool_result , and text_editor_code_execution_tool_result content block types

• Add tool_search_tool_result content block type

• Add tool_use_id and content properties on CreateResponseContent for all server tool result blocks

• Add container field on CreateResponse for code execution sandbox persistence

• Add citations array on text content blocks, supporting all 5 citation location types: char_location , page_location , content_block_location , web_search_result_location , search_result_location

• Add citations_delta streaming support with citation property on CreateStreamedResponseDelta

• Add webFetchRequests , codeExecutionRequests , and toolSearchRequests to CreateResponseUsageServerToolUse

• Add streaming support for server_tool_use and server tool result blocks in CreateStreamedResponseContentBlockStart

Full Changelog : 1.3.3...1.4.0

• updated model names for consistency

Full Changelog: 1.3.2...1.3.3

• updated model names for consistency

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.3.2...1.3.3

• updated model names for consistency

Full Changelog : 1.3.2...1.3.3

• add batch response return type to client fake

Full Changelog: 1.3.1...1.3.2

• add batch response return type to client fake

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.3.1...1.3.2

• add batch response return type to client fake

Full Changelog : 1.3.1...1.3.2

What's Changed

Added

• Add dedicated inputTokenLimit and outputTokenLimit rate limit properties to MetaInformation
• Add cache_creation breakdown to usage (CreateResponseUsageCacheCreation with ephemeral5mInputTokens and ephemeral1hInputTokens)
• Add serviceTier field to usage (standard, priority, or batch)
• Add serverToolUse field to usage (CreateResponseUsageServerToolUse with webSearchRequests)

Improved

• Input/output token rate limit headers are now parsed as dedicated properties instead of falling into the generic custom bucket
• Usage objects now capture all fields returned by the API instead of silently dropping extended fields

Full Changelog: 1.3.0...1.3.1

What's Changed

Added

• Add dedicated inputTokenLimit and outputTokenLimit rate limit properties to MetaInformation
• Add cache_creation breakdown to usage (CreateResponseUsageCacheCreation with ephemeral5mInputTokens and ephemeral1hInputTokens)
• Add serviceTier field to usage (standard, priority, or batch)
• Add serverToolUse field to usage (CreateResponseUsageServerToolUse with webSearchRequests)

Improved

• Input/output token rate limit headers are now parsed as dedicated properties instead of falling into the generic custom bucket
• Usage objects now capture all fields returned by the API instead of silently dropping extended fields

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.3.0...1.3.1

What's Changed

Added

• Add dedicated inputTokenLimit and outputTokenLimit rate limit properties to MetaInformation

• Add cache_creation breakdown to usage ( CreateResponseUsageCacheCreation with ephemeral5mInputTokens and ephemeral1hInputTokens )

• Add serviceTier field to usage (standard, priority, or batch)

• Add serverToolUse field to usage ( CreateResponseUsageServerToolUse with webSearchRequests )

Improved

• Input/output token rate limit headers are now parsed as dedicated properties instead of falling into the generic custom bucket

• Usage objects now capture all fields returned by the API instead of silently dropping extended fields

Full Changelog : 1.3.0...1.3.1

What's Changed

Added

• Add Message Batches support — create, retrieve, list, cancel, delete, and stream results
• Add BatchResponse, BatchListResponse, DeletedBatchResponse, and BatchResultResponse response DTOs
• Add BatchIndividualResponse, BatchResult, and BatchResultError for parsing JSONL batch results
• Add BatchResponseRequestCounts for tracking processing, succeeded, errored, canceled, and expired counts
• Add BatchResultResponse::fake() for consumer testing with ClientFake

Full Changelog: 1.2.3...1.3.0

What's Changed

Added

• Add Message Batches support — create, retrieve, list, cancel, delete, and stream results
• Add BatchResponse, BatchListResponse, DeletedBatchResponse, and BatchResultResponse response DTOs
• Add BatchIndividualResponse, BatchResult, and BatchResultError for parsing JSONL batch results
• Add BatchResponseRequestCounts for tracking processing, succeeded, errored, canceled, and expired counts
• Add BatchResultResponse::fake() for consumer testing with ClientFake
• Add dedicated inputTokenLimit and outputTokenLimit rate limit properties to MetaInformation
• Add cache_creation breakdown to usage (CreateResponseUsageCacheCreation with ephemeral5mInputTokens and ephemeral1hInputTokens)
• Add serviceTier field to usage (standard, priority, or batch)
• Add serverToolUse field to usage (CreateResponseUsageServerToolUse with webSearchRequests)

Improved

• Input/output token rate limit headers are now parsed as dedicated properties instead of falling into the generic custom bucket
• Usage objects now capture all fields returned by the API instead of silently dropping extended fields

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.2.3...1.3.0

What's Changed

Added

• Add Message Batches support — create, retrieve, list, cancel, delete, and stream results

• Add BatchResponse , BatchListResponse , DeletedBatchResponse , and BatchResultResponse response DTOs

• Add BatchIndividualResponse , BatchResult , and BatchResultError for parsing JSONL batch results

• Add BatchResponseRequestCounts for tracking processing, succeeded, errored, canceled, and expired counts

• Add BatchResultResponse::fake() for consumer testing with ClientFake

• Add dedicated inputTokenLimit and outputTokenLimit rate limit properties to MetaInformation

• Add cache_creation breakdown to usage ( CreateResponseUsageCacheCreation with ephemeral5mInputTokens and ephemeral1hInputTokens )

• Add serviceTier field to usage (standard, priority, or batch)

• Add serverToolUse field to usage ( CreateResponseUsageServerToolUse with webSearchRequests )

Improved

• Input/output token rate limit headers are now parsed as dedicated properties instead of falling into the generic custom bucket

• Usage objects now capture all fields returned by the API instead of silently dropping extended fields

Full Changelog : 1.2.3...1.3.0

What's Changed

• add count tokens support
• add models list and retrieve support

Full Changelog: 1.2.2...1.2.3

What's Changed

• add count tokens support
• add models list and retrieve support

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.2.2...1.2.3

What's Changed

• add count tokens support

• add models list and retrieve support

Full Changelog : 1.2.2...1.2.3

What's Changed

• added thinking support
• add anthropic-version header to the factory

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.2.1...1.2.2

What's Changed

• added thinking support

• add anthropic-version header to the factory

Full Changelog : 1.2.1...1.2.2

What's Changed

Fixed

• Add missing cache_creation_input_tokens and cache_read_input_tokens to CreateStreamedResponseUsage

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.2.0...1.2.1

What's Changed

Fixed

• Add missing cache_creation_input_tokens and cache_read_input_tokens to CreateStreamedResponseUsage

Full Changelog : 1.2.0...1.2.1

What's Changed

Added

• Add RateLimitException for HTTP 429 responses (extends ErrorException — no breaking changes)
• Add PSR-7 response access on ErrorException and UnserializableResponse via $e->response
• Add MetaInformationCustom for capturing non-standard response headers via $meta->custom
• Add OverrideStrategy enum for controlling fake() override merging behavior
• Add addHeader() method to TransporterContract for runtime header injection

Improved

• Preserve API error message on rate limit responses instead of using a generic message
• Use (string) $response->getBody() instead of getContents() for reliable stream reading
• Add JSON_UNESCAPED_UNICODE flag for proper Unicode handling in request payloads
• Simplify Fakeable trait using array_replace_recursive
• Fix namespace resolution in Fakeable str_replace call
• Parameterize Streamable trait methods for flexibility

Upgrading

See UPGRADING.md for details. All changes are backwards compatible.

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.1.0...1.2.0

What's Changed

Added

• Add RateLimitException for HTTP 429 responses (extends ErrorException — no breaking changes)

• Add PSR-7 response access on ErrorException and UnserializableResponse via $e->response

• Add MetaInformationCustom for capturing non-standard response headers via $meta->custom

• Add OverrideStrategy enum for controlling fake() override merging behavior

• Add addHeader() method to TransporterContract for runtime header injection

Improved

• Preserve API error message on rate limit responses instead of using a generic message

• Use (string) $response->getBody() instead of getContents() for reliable stream reading

• Add JSON_UNESCAPED_UNICODE flag for proper Unicode handling in request payloads

• Simplify Fakeable trait using array_replace_recursive

• Fix namespace resolution in Fakeable str_replace call

• Parameterize Streamable trait methods for flexibility

Upgrading

See UPGRADING.md for details. All changes are backwards compatible.

Full Changelog : 1.1.0...1.2.0

What's Changed

• Add tool use (function calling) support by @mozex in https://github.com/mozex/anthropic-php/pull/6
• Add cached tokens usage by @mozex in https://github.com/mozex/anthropic-php/pull/7

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.0.3...1.1.0

What's Changed

• Add tool use (function calling) support by @mozex in #6

• Add cached tokens usage by @mozex in #7

Full Changelog : 1.0.3...1.1.0

• bump dependencies
• add php 8.4 to workflow
• add status code to error exceptions

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.0.2...1.0.3

• bump dependencies

• add php 8.4 to workflow

• add status code to error exceptions

Full Changelog : 1.0.2...1.0.3

What's Changed

• Bump dependencies
• Refactoring

Full Changelog: https://github.com/mozex/anthropic-php/compare/1.0.1...1.0.2

What's Changed

• Bump dependencies

• Refactoring

Full Changelog : 1.0.1...1.0.2

• fix client fake types

Initial Release

Initial Release

Scroll to top