ruby_llm 1.8.2 → 1.9.1

data/lib/ruby_llm/aliases.json

@@ -8,16 +8,17 @@
     "openrouter": "anthropic/claude-3.5-haiku",
     "bedrock": "anthropic.claude-3-5-haiku-20241022-v1:0"
   },
-  "claude-3-5-sonnet": {
-    "anthropic": "claude-3-5-sonnet-20241022",
-    "openrouter": "anthropic/claude-3.5-sonnet",
-    "bedrock": "anthropic.claude-3-5-sonnet-20240620-v1:0:200k"
+  "claude-3-5-haiku-latest": {
+    "anthropic": "claude-3-5-haiku-latest"
   },
   "claude-3-7-sonnet": {
     "anthropic": "claude-3-7-sonnet-20250219",
     "openrouter": "anthropic/claude-3.7-sonnet",
     "bedrock": "us.anthropic.claude-3-7-sonnet-20250219-v1:0"
   },
+  "claude-3-7-sonnet-latest": {
+    "anthropic": "claude-3-7-sonnet-latest"
+  },
   "claude-3-haiku": {
     "anthropic": "claude-3-haiku-20240307",
     "openrouter": "anthropic/claude-3-haiku",
@@ -31,11 +32,19 @@
   "claude-3-sonnet": {
     "bedrock": "anthropic.claude-3-sonnet-20240229-v1:0"
   },
+  "claude-haiku-4-5": {
+    "anthropic": "claude-haiku-4-5-20251001",
+    "openrouter": "anthropic/claude-haiku-4.5",
+    "bedrock": "us.anthropic.claude-haiku-4-5-20251001-v1:0"
+  },
   "claude-opus-4": {
     "anthropic": "claude-opus-4-20250514",
     "openrouter": "anthropic/claude-opus-4",
     "bedrock": "us.anthropic.claude-opus-4-1-20250805-v1:0"
   },
+  "claude-opus-4-0": {
+    "anthropic": "claude-opus-4-0"
+  },
   "claude-opus-4-1": {
     "anthropic": "claude-opus-4-1-20250805",
     "openrouter": "anthropic/claude-opus-4.1",
@@ -46,30 +55,18 @@
     "openrouter": "anthropic/claude-sonnet-4",
     "bedrock": "us.anthropic.claude-sonnet-4-20250514-v1:0"
   },
+  "claude-sonnet-4-0": {
+    "anthropic": "claude-sonnet-4-0"
+  },
+  "claude-sonnet-4-5": {
+    "anthropic": "claude-sonnet-4-5-20250929",
+    "openrouter": "anthropic/claude-sonnet-4.5",
+    "bedrock": "us.anthropic.claude-sonnet-4-5-20250929-v1:0"
+  },
   "deepseek-chat": {
     "deepseek": "deepseek-chat",
     "openrouter": "deepseek/deepseek-chat"
   },
-  "gemini-1.5-flash": {
-    "gemini": "gemini-1.5-flash",
-    "vertexai": "gemini-1.5-flash"
-  },
-  "gemini-1.5-flash-002": {
-    "gemini": "gemini-1.5-flash-002",
-    "vertexai": "gemini-1.5-flash-002"
-  },
-  "gemini-1.5-flash-8b": {
-    "gemini": "gemini-1.5-flash-8b",
-    "vertexai": "gemini-1.5-flash-8b"
-  },
-  "gemini-1.5-pro": {
-    "gemini": "gemini-1.5-pro",
-    "vertexai": "gemini-1.5-pro"
-  },
-  "gemini-1.5-pro-002": {
-    "gemini": "gemini-1.5-pro-002",
-    "vertexai": "gemini-1.5-pro-002"
-  },
   "gemini-2.0-flash": {
     "gemini": "gemini-2.0-flash",
     "vertexai": "gemini-2.0-flash"
@@ -93,6 +90,10 @@
     "openrouter": "google/gemini-2.5-flash",
     "vertexai": "gemini-2.5-flash"
   },
+  "gemini-2.5-flash-image": {
+    "gemini": "gemini-2.5-flash-image",
+    "openrouter": "google/gemini-2.5-flash-image"
+  },
   "gemini-2.5-flash-image-preview": {
     "gemini": "gemini-2.5-flash-image-preview",
     "openrouter": "google/gemini-2.5-flash-image-preview"
@@ -106,6 +107,14 @@
     "gemini": "gemini-2.5-flash-lite-preview-06-17",
     "openrouter": "google/gemini-2.5-flash-lite-preview-06-17"
   },
+  "gemini-2.5-flash-lite-preview-09-2025": {
+    "gemini": "gemini-2.5-flash-lite-preview-09-2025",
+    "openrouter": "google/gemini-2.5-flash-lite-preview-09-2025"
+  },
+  "gemini-2.5-flash-preview-09-2025": {
+    "gemini": "gemini-2.5-flash-preview-09-2025",
+    "openrouter": "google/gemini-2.5-flash-preview-09-2025"
+  },
   "gemini-2.5-pro": {
     "gemini": "gemini-2.5-pro",
     "openrouter": "google/gemini-2.5-pro",
@@ -219,6 +228,10 @@
     "openai": "gpt-5",
     "openrouter": "openai/gpt-5"
   },
+  "gpt-5-codex": {
+    "openai": "gpt-5-codex",
+    "openrouter": "openai/gpt-5-codex"
+  },
   "gpt-5-mini": {
     "openai": "gpt-5-mini",
     "openrouter": "openai/gpt-5-mini"
@@ -227,18 +240,26 @@
     "openai": "gpt-5-nano",
     "openrouter": "openai/gpt-5-nano"
   },
+  "gpt-5-pro": {
+    "openai": "gpt-5-pro",
+    "openrouter": "openai/gpt-5-pro"
+  },
+  "gpt-oss-120b": {
+    "openai": "gpt-oss-120b",
+    "openrouter": "openai/gpt-oss-120b"
+  },
+  "gpt-oss-20b": {
+    "openai": "gpt-oss-20b",
+    "openrouter": "openai/gpt-oss-20b"
+  },
+  "imagen-4.0-generate-001": {
+    "gemini": "imagen-4.0-generate-001",
+    "vertexai": "imagen-4.0-generate-001"
+  },
   "o1": {
     "openai": "o1",
     "openrouter": "openai/o1"
   },
-  "o1-mini": {
-    "openai": "o1-mini",
-    "openrouter": "openai/o1-mini"
-  },
-  "o1-mini-2024-09-12": {
-    "openai": "o1-mini-2024-09-12",
-    "openrouter": "openai/o1-mini-2024-09-12"
-  },
   "o1-pro": {
     "openai": "o1-pro",
     "openrouter": "openai/o1-pro"
@@ -247,6 +268,10 @@
     "openai": "o3",
     "openrouter": "openai/o3"
   },
+  "o3-deep-research": {
+    "openai": "o3-deep-research",
+    "openrouter": "openai/o3-deep-research"
+  },
   "o3-mini": {
     "openai": "o3-mini",
     "openrouter": "openai/o3-mini"
@@ -259,6 +284,10 @@
     "openai": "o4-mini",
     "openrouter": "openai/o4-mini"
   },
+  "o4-mini-deep-research": {
+    "openai": "o4-mini-deep-research",
+    "openrouter": "openai/o4-mini-deep-research"
+  },
   "text-embedding-004": {
     "gemini": "text-embedding-004",
     "vertexai": "text-embedding-004"

data/lib/ruby_llm/attachment.rb

@@ -7,17 +7,8 @@ module RubyLLM
 
     def initialize(source, filename: nil)
       @source = source
-      if url?
-        @source = URI source
-        @filename = filename || File.basename(@source.path).to_s
-      elsif path?
-        @source = Pathname.new source
-        @filename = filename || @source.basename.to_s
-      elsif active_storage?
-        @filename = filename || extract_filename_from_active_storage
-      else
-        @filename = filename
-      end
+      @source = source_type_cast
+      @filename = filename || source_filename
 
       determine_mime_type
     end
@@ -65,6 +56,14 @@ module RubyLLM
       Base64.strict_encode64(content)
     end
 
+    def save(path)
+      return unless io_like?
+
+      File.open(path, 'w') do |f|
+        f.puts(@source.read)
+      end
+    end
+
     def for_llm
       case type
       when :text
@@ -158,6 +157,38 @@ module RubyLLM
                  end
     end
 
+    def source_type_cast
+      if url?
+        URI(@source)
+      elsif path?
+        Pathname.new(@source)
+      else
+        @source
+      end
+    end
+
+    def source_filename
+      if url?
+        File.basename(@source.path).to_s
+      elsif path?
+        @source.basename.to_s
+      elsif io_like?
+        extract_filename_from_io
+      elsif active_storage?
+        extract_filename_from_active_storage
+      end
+    end
+
+    def extract_filename_from_io
+      if defined?(ActionDispatch::Http::UploadedFile) && @source.is_a?(ActionDispatch::Http::UploadedFile)
+        @source.original_filename.to_s
+      elsif @source.respond_to?(:path)
+        File.basename(@source.path).to_s
+      else
+        'attachment'
+      end
+    end
+
     def extract_filename_from_active_storage # rubocop:disable Metrics/PerceivedComplexity
       return 'attachment' unless defined?(ActiveStorage)
 

data/lib/ruby_llm/chat.rb

@@ -31,7 +31,7 @@ module RubyLLM
     end
 
     def ask(message = nil, with: nil, &)
-      add_message role: :user, content: Content.new(message, with)
+      add_message role: :user, content: build_content(message, with)
       complete(&)
     end
 
@@ -193,7 +193,8 @@ module RubyLLM
         @on[:tool_call]&.call(tool_call)
         result = execute_tool tool_call
         @on[:tool_result]&.call(result)
-        content = result.is_a?(Content) ? result : result.to_s
+        tool_payload = result.is_a?(Tool::Halt) ? result.content : result
+        content = content_like?(tool_payload) ? tool_payload : tool_payload.to_s
         message = add_message role: :tool, content:, tool_call_id: tool_call.id
         @on[:end_message]&.call(message)
 
@@ -208,5 +209,15 @@ module RubyLLM
       args = tool_call.arguments
       tool.call(args)
     end
+
+    def build_content(message, attachments)
+      return message if content_like?(message)
+
+      Content.new(message, attachments)
+    end
+
+    def content_like?(object)
+      object.is_a?(Content) || object.is_a?(Content::Raw)
+    end
   end
 end

data/lib/ruby_llm/configuration.rb

@@ -10,6 +10,7 @@ module RubyLLM
                   :openai_use_system_role,
                   :anthropic_api_key,
                   :gemini_api_key,
+                  :gemini_api_base,
                   :vertexai_project_id,
                   :vertexai_location,
                   :deepseek_api_key,
@@ -28,7 +29,9 @@ module RubyLLM
                   :default_embedding_model,
                   :default_moderation_model,
                   :default_image_model,
+                  :default_transcription_model,
                   # Model registry
+                  :model_registry_file,
                   :model_registry_class,
                   # Rails integration
                   :use_new_acts_as,
@@ -46,7 +49,7 @@ module RubyLLM
                   :log_stream_debug
 
     def initialize
-      @request_timeout = 120
+      @request_timeout = 300
       @max_retries = 3
       @retry_interval = 0.1
       @retry_backoff_factor = 2
@@ -57,7 +60,9 @@ module RubyLLM
       @default_embedding_model = 'text-embedding-3-small'
       @default_moderation_model = 'omni-moderation-latest'
       @default_image_model = 'gpt-image-1'
+      @default_transcription_model = 'whisper-1'
 
+      @model_registry_file = File.expand_path('models.json', __dir__)
       @model_registry_class = 'Model'
       @use_new_acts_as = false
 

data/lib/ruby_llm/connection.rb

@@ -34,8 +34,7 @@ module RubyLLM
     end
 
     def post(url, payload, &)
-      body = payload.is_a?(Hash) ? JSON.generate(payload, ascii_only: false) : payload
-      @connection.post url, body do |req|
+      @connection.post url, payload do |req|
         req.headers.merge! @provider.headers if @provider.respond_to?(:headers)
         yield req if block_given?
       end
@@ -66,7 +65,7 @@ module RubyLLM
                        errors: true,
                        headers: false,
                        log_level: :debug do |logger|
-        logger.filter(%r{[A-Za-z0-9+/=]{100,}}, 'data":"[BASE64 DATA]"')
+        logger.filter(%r{[A-Za-z0-9+/=]{100,}}, '[BASE64 DATA]')
         logger.filter(/[-\d.e,\s]{100,}/, '[EMBEDDINGS ARRAY]')
       end
     end
@@ -83,6 +82,7 @@ module RubyLLM
     end
 
     def setup_middleware(faraday)
+      faraday.request :multipart
       faraday.request :json
       faraday.response :json
       faraday.adapter :net_http

data/lib/ruby_llm/content.rb

@@ -48,3 +48,26 @@ module RubyLLM
     end
   end
 end
+
+module RubyLLM
+  class Content
+    # Represents provider-specific payloads that should bypass RubyLLM formatting.
+    class Raw
+      attr_reader :value
+
+      def initialize(value)
+        raise ArgumentError, 'Raw content payload cannot be nil' if value.nil?
+
+        @value = value
+      end
+
+      def format
+        @value
+      end
+
+      def to_h
+        @value
+      end
+    end
+  end
+end

data/lib/ruby_llm/message.rb

@@ -5,17 +5,20 @@ module RubyLLM
   class Message
     ROLES = %i[system user assistant tool].freeze
 
-    attr_reader :role, :tool_calls, :tool_call_id, :input_tokens, :output_tokens, :model_id, :raw
+    attr_reader :role, :model_id, :tool_calls, :tool_call_id, :input_tokens, :output_tokens,
+                :cached_tokens, :cache_creation_tokens, :raw
     attr_writer :content
 
     def initialize(options = {})
       @role = options.fetch(:role).to_sym
       @content = normalize_content(options.fetch(:content))
+      @model_id = options[:model_id]
       @tool_calls = options[:tool_calls]
+      @tool_call_id = options[:tool_call_id]
       @input_tokens = options[:input_tokens]
       @output_tokens = options[:output_tokens]
-      @model_id = options[:model_id]
-      @tool_call_id = options[:tool_call_id]
+      @cached_tokens = options[:cached_tokens]
+      @cache_creation_tokens = options[:cache_creation_tokens]
       @raw = options[:raw]
 
       ensure_valid_role
@@ -45,11 +48,13 @@ module RubyLLM
       {
         role: role,
         content: content,
+        model_id: model_id,
         tool_calls: tool_calls,
         tool_call_id: tool_call_id,
         input_tokens: input_tokens,
         output_tokens: output_tokens,
-        model_id: model_id
+        cached_tokens: cached_tokens,
+        cache_creation_tokens: cache_creation_tokens
       }.compact
     end
 

data/lib/ruby_llm/model/info.rb

@@ -72,6 +72,10 @@ module RubyLLM
         pricing.text_tokens.output
       end
 
+      def provider_class
+        RubyLLM::Provider.resolve provider
+      end
+
       def type # rubocop:disable Metrics/PerceivedComplexity
         if modalities.output.include?('embeddings') && !modalities.output.include?('text')
           'embedding'