ruby_llm 1.3.2beta1 → 1.5.0

data/lib/ruby_llm/providers/mistral.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    # Mistral API integration.
+    module Mistral
+      extend OpenAI
+      extend Mistral::Chat
+      extend Mistral::Models
+      extend Mistral::Embeddings
+
+      module_function
+
+      def api_base(_config)
+        'https://api.mistral.ai/v1'
+      end
+
+      def headers(config)
+        {
+          'Authorization' => "Bearer #{config.mistral_api_key}"
+        }
+      end
+
+      def capabilities
+        Mistral::Capabilities
+      end
+
+      def slug
+        'mistral'
+      end
+
+      def configuration_requirements
+        %i[mistral_api_key]
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/ollama/media.rb

@@ -10,6 +10,8 @@ module RubyLLM
         module_function
 
         def format_content(content)
+          # Convert Hash/Array back to JSON string for API
+          return content.to_json if content.is_a?(Hash) || content.is_a?(Array)
           return content unless content.is_a?(Content)
 
           parts = []

data/lib/ruby_llm/providers/openai/chat.rb

@@ -11,7 +11,7 @@ module RubyLLM
 
         module_function
 
-        def render_payload(messages, tools:, temperature:, model:, stream: false)
+        def render_payload(messages, tools:, temperature:, model:, stream: false, schema: nil) # rubocop:disable Metrics/ParameterLists
           payload = {
             model: model,
             messages: format_messages(messages),
@@ -26,6 +26,20 @@ module RubyLLM
             payload[:tool_choice] = 'auto'
           end
 
+          if schema
+            # Use strict mode from schema if specified, default to true
+            strict = schema[:strict] != false
+
+            payload[:response_format] = {
+              type: 'json_schema',
+              json_schema: {
+                name: 'response',
+                schema: schema,
+                strict: strict
+              }
+            }
+          end
+
           payload[:stream_options] = { include_usage: true } if stream
           payload
         end
@@ -45,7 +59,8 @@ module RubyLLM
             tool_calls: parse_tool_calls(message_data['tool_calls']),
             input_tokens: data['usage']['prompt_tokens'],
             output_tokens: data['usage']['completion_tokens'],
-            model_id: data['model']
+            model_id: data['model'],
+            raw: response
           )
         end
 

data/lib/ruby_llm/providers/openai/media.rb

@@ -8,6 +8,8 @@ module RubyLLM
         module_function
 
         def format_content(content)
+          # Convert Hash/Array back to JSON string for API
+          return content.to_json if content.is_a?(Hash) || content.is_a?(Array)
           return content unless content.is_a?(Content)
 
           parts = []

data/lib/ruby_llm/providers/openai/streaming.rb

@@ -21,6 +21,20 @@ module RubyLLM
             output_tokens: data.dig('usage', 'completion_tokens')
           )
         end
+
+        def parse_streaming_error(data)
+          error_data = JSON.parse(data)
+          return unless error_data['error']
+
+          case error_data.dig('error', 'type')
+          when 'server_error'
+            [500, error_data['error']['message']]
+          when 'rate_limit_exceeded', 'insufficient_quota'
+            [429, error_data['error']['message']]
+          else
+            [400, error_data['error']['message']]
+          end
+        end
       end
     end
   end

data/lib/ruby_llm/providers/perplexity/capabilities.rb

@@ -0,0 +1,167 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module Perplexity
+      # Determines capabilities and pricing for Perplexity models
+      module Capabilities
+        module_function
+
+        # Returns the context window size for the given model ID
+        # @param model_id [String] the model identifier
+        # @return [Integer] the context window size in tokens
+        def context_window_for(model_id)
+          case model_id
+          when /sonar-pro/ then 200_000
+          else 128_000
+          end
+        end
+
+        # Returns the maximum number of tokens that can be generated
+        # @param model_id [String] the model identifier
+        # @return [Integer] the maximum number of tokens
+        def max_tokens_for(model_id)
+          case model_id
+          when /sonar-(?:pro|reasoning-pro)/ then 8_192
+          else 4_096
+          end
+        end
+
+        # Returns the price per million tokens for input
+        # @param model_id [String] the model identifier
+        # @return [Float] the price per million tokens in USD
+        def input_price_for(model_id)
+          PRICES.dig(model_family(model_id), :input) || 1.0
+        end
+
+        # Returns the price per million tokens for output
+        # @param model_id [String] the model identifier
+        # @return [Float] the price per million tokens in USD
+        def output_price_for(model_id)
+          PRICES.dig(model_family(model_id), :output) || 1.0
+        end
+
+        # Determines if the model supports vision capabilities
+        # @param model_id [String] the model identifier
+        # @return [Boolean] true if the model supports vision
+        def supports_vision?(model_id)
+          case model_id
+          when /sonar-reasoning-pro/, /sonar-reasoning/, /sonar-pro/, /sonar/ then true
+          else false
+          end
+        end
+
+        # Determines if the model supports function calling
+        # @param model_id [String] the model identifier
+        # @return [Boolean] true if the model supports functions
+        def supports_functions?(_model_id)
+          false
+        end
+
+        # Determines if the model supports JSON mode
+        def supports_json_mode?(_model_id)
+          true
+        end
+
+        # Formats the model ID into a human-readable display name
+        # @param model_id [String] the model identifier
+        # @return [String] the formatted display name
+        def format_display_name(model_id)
+          case model_id
+          when 'sonar' then 'Sonar'
+          when 'sonar-pro' then 'Sonar Pro'
+          when 'sonar-reasoning' then 'Sonar Reasoning'
+          when 'sonar-reasoning-pro' then 'Sonar Reasoning Pro'
+          when 'sonar-deep-research' then 'Sonar Deep Research'
+          else
+            model_id.split('-')
+                    .map(&:capitalize)
+                    .join(' ')
+          end
+        end
+
+        # Returns the model type
+        # @param model_id [String] the model identifier
+        # @return [String] the model type (e.g., 'chat')
+        def model_type(_model_id)
+          'chat'
+        end
+
+        # Returns the model family
+        # @param model_id [String] the model identifier
+        # @return [Symbol] the model family
+        def model_family(model_id)
+          case model_id
+          when 'sonar' then :sonar
+          when 'sonar-pro' then :sonar_pro
+          when 'sonar-reasoning' then :sonar_reasoning
+          when 'sonar-reasoning-pro' then :sonar_reasoning_pro
+          when 'sonar-deep-research' then :sonar_deep_research
+          else :unknown
+          end
+        end
+
+        def modalities_for(_model_id)
+          {
+            input: ['text'],
+            output: ['text']
+          }
+        end
+
+        def capabilities_for(model_id)
+          capabilities = %w[streaming json_mode]
+          capabilities << 'vision' if supports_vision?(model_id)
+          capabilities
+        end
+
+        def pricing_for(model_id)
+          family = model_family(model_id)
+          prices = PRICES.fetch(family, { input: 1.0, output: 1.0 })
+
+          standard_pricing = {
+            input_per_million: prices[:input],
+            output_per_million: prices[:output]
+          }
+
+          # Add special pricing if available
+          standard_pricing[:citation_per_million] = prices[:citation] if prices[:citation]
+          standard_pricing[:reasoning_per_million] = prices[:reasoning] if prices[:reasoning]
+          standard_pricing[:search_per_thousand] = prices[:search_queries] if prices[:search_queries]
+
+          {
+            text_tokens: {
+              standard: standard_pricing
+            }
+          }
+        end
+
+        # Pricing information for Perplexity models (USD per 1M tokens)
+        PRICES = {
+          sonar: {
+            input: 1.0,
+            output: 1.0
+          },
+          sonar_pro: {
+            input: 3.0,
+            output: 15.0
+          },
+          sonar_reasoning: {
+            input: 1.0,
+            output: 5.0
+          },
+          sonar_reasoning_pro: {
+            input: 2.0,
+            output: 8.0
+          },
+          sonar_deep_research: {
+            input: 2.0,
+            output: 8.0,
+            citation: 2.0,
+            reasoning: 3.0,
+            search_queries: 5.0
+          }
+        }.freeze
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/perplexity/chat.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module Perplexity
+      # Chat formatting for Perplexity provider
+      module Chat
+        module_function
+
+        def format_role(role)
+          # Perplexity doesn't use the new OpenAI convention for system prompts
+          role.to_s
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/perplexity/models.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    module Perplexity
+      # Models methods of the Perplexity API integration
+      module Models
+        def list_models(**)
+          slug = 'perplexity'
+          capabilities = Perplexity::Capabilities
+          parse_list_models_response(nil, slug, capabilities)
+        end
+
+        def parse_list_models_response(_response, slug, capabilities)
+          [
+            create_model_info('sonar', slug, capabilities),
+            create_model_info('sonar-pro', slug, capabilities),
+            create_model_info('sonar-reasoning', slug, capabilities),
+            create_model_info('sonar-reasoning-pro', slug, capabilities),
+            create_model_info('sonar-deep-research', slug, capabilities)
+          ]
+        end
+
+        def create_model_info(id, slug, capabilities)
+          Model::Info.new(
+            id: id,
+            name: capabilities.format_display_name(id),
+            provider: slug,
+            family: capabilities.model_family(id).to_s,
+            created_at: Time.now,
+            context_window: capabilities.context_window_for(id),
+            max_output_tokens: capabilities.max_tokens_for(id),
+            modalities: capabilities.modalities_for(id),
+            capabilities: capabilities.capabilities_for(id),
+            pricing: capabilities.pricing_for(id),
+            metadata: {}
+          )
+        end
+      end
+    end
+  end
+end

data/lib/ruby_llm/providers/perplexity.rb

@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+module RubyLLM
+  module Providers
+    # Perplexity API integration.
+    module Perplexity
+      extend OpenAI
+      extend Perplexity::Chat
+      extend Perplexity::Models
+
+      module_function
+
+      def api_base(_config)
+        'https://api.perplexity.ai'
+      end
+
+      def headers(config)
+        {
+          'Authorization' => "Bearer #{config.perplexity_api_key}",
+          'Content-Type' => 'application/json'
+        }
+      end
+
+      def capabilities
+        Perplexity::Capabilities
+      end
+
+      def slug
+        'perplexity'
+      end
+
+      def configuration_requirements
+        %i[perplexity_api_key]
+      end
+
+      def parse_error(response)
+        body = response.body
+        return if body.empty?
+
+        # If response is HTML (Perplexity returns HTML for auth errors)
+        if body.include?('<html>') && body.include?('<title>')
+          # Extract title content
+          title_match = body.match(%r{<title>(.+?)</title>})
+          if title_match
+            # Clean up the title - remove status code if present
+            message = title_match[1]
+            message = message.sub(/^\d+\s+/, '') # Remove leading digits and space
+            return message
+          end
+        end
+
+        # Fall back to parent's implementation
+        super
+      end
+    end
+  end
+end

data/lib/ruby_llm/railtie.rb

@@ -8,5 +8,10 @@ module RubyLLM
         include RubyLLM::ActiveRecord::ActsAs
       end
     end
+
+    # Register generators
+    generators do
+      require 'generators/ruby_llm/install_generator'
+    end
   end
 end

data/lib/ruby_llm/stream_accumulator.rb

@@ -29,14 +29,15 @@ module RubyLLM
       RubyLLM.logger.debug inspect
     end
 
-    def to_message
+    def to_message(response)
       Message.new(
         role: :assistant,
         content: content.empty? ? nil : content,
         model_id: model_id,
         tool_calls: tool_calls_from_stream,
         input_tokens: @input_tokens.positive? ? @input_tokens : nil,
-        output_tokens: @output_tokens.positive? ? @output_tokens : nil
+        output_tokens: @output_tokens.positive? ? @output_tokens : nil,
+        raw: response
       )
     end
 

data/lib/ruby_llm/streaming.rb

@@ -11,7 +11,7 @@ module RubyLLM
     def stream_response(connection, payload, &block)
       accumulator = StreamAccumulator.new
 
-      connection.post stream_url, payload do |req|
+      response = connection.post stream_url, payload do |req|
         if req.options.respond_to?(:on_data)
           # Handle Faraday 2.x streaming with on_data method
           req.options.on_data = handle_stream do |chunk|
@@ -27,7 +27,7 @@ module RubyLLM
         end
       end
 
-      accumulator.to_message
+      accumulator.to_message(response)
     end
 
     def handle_stream(&block)
@@ -55,13 +55,13 @@ module RubyLLM
       end
     end
 
-    def process_stream_chunk(chunk, parser, _env, &)
+    def process_stream_chunk(chunk, parser, env, &)
       RubyLLM.logger.debug "Received chunk: #{chunk}"
 
       if error_chunk?(chunk)
-        handle_error_chunk(chunk, nil)
+        handle_error_chunk(chunk, env)
       else
-        yield handle_sse(chunk, parser, nil, &)
+        yield handle_sse(chunk, parser, env, &)
       end
     end
 
@@ -88,7 +88,16 @@ module RubyLLM
     def handle_error_chunk(chunk, env)
       error_data = chunk.split("\n")[1].delete_prefix('data: ')
       status, _message = parse_streaming_error(error_data)
-      error_response = env.merge(body: JSON.parse(error_data), status: status)
+      parsed_data = JSON.parse(error_data)
+
+      # Create a response-like object that works for both Faraday v1 and v2
+      error_response = if env
+                         env.merge(body: parsed_data, status: status)
+                       else
+                         # For Faraday v1, create a simple object that responds to .status and .body
+                         Struct.new(:body, :status).new(parsed_data, status)
+                       end
+
       ErrorMiddleware.parse_error(provider: self, response: error_response)
     rescue JSON::ParserError => e
       RubyLLM.logger.debug "Failed to parse error chunk: #{e.message}"
@@ -122,7 +131,16 @@ module RubyLLM
 
     def handle_error_event(data, env)
       status, _message = parse_streaming_error(data)
-      error_response = env.merge(body: JSON.parse(data), status: status)
+      parsed_data = JSON.parse(data)
+
+      # Create a response-like object that works for both Faraday v1 and v2
+      error_response = if env
+                         env.merge(body: parsed_data, status: status)
+                       else
+                         # For Faraday v1, create a simple object that responds to .status and .body
+                         Struct.new(:body, :status).new(parsed_data, status)
+                       end
+
       ErrorMiddleware.parse_error(provider: self, response: error_response)
     rescue JSON::ParserError => e
       RubyLLM.logger.debug "Failed to parse error event: #{e.message}"

data/lib/ruby_llm/utils.rb

@@ -23,5 +23,15 @@ module RubyLLM
         Array(item)
       end
     end
+
+    def deep_merge(params, payload)
+      params.merge(payload) do |_key, params_value, payload_value|
+        if params_value.is_a?(Hash) && payload_value.is_a?(Hash)
+          deep_merge(params_value, payload_value)
+        else
+          payload_value
+        end
+      end
+    end
   end
 end

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '1.3.2beta1'
+  VERSION = '1.5.0'
 end

data/lib/ruby_llm.rb

@@ -16,13 +16,17 @@ loader.inflector.inflect(
   'openai' => 'OpenAI',
   'api' => 'API',
   'deepseek' => 'DeepSeek',
+  'perplexity' => 'Perplexity',
   'bedrock' => 'Bedrock',
   'openrouter' => 'OpenRouter',
+  'gpustack' => 'GPUStack',
+  'mistral' => 'Mistral',
   'pdf' => 'PDF'
 )
 loader.ignore("#{__dir__}/tasks")
 loader.ignore("#{__dir__}/ruby_llm/railtie")
 loader.ignore("#{__dir__}/ruby_llm/active_record")
+loader.ignore("#{__dir__}/generators")
 loader.setup
 
 # A delightful Ruby interface to modern AI language models.
@@ -76,13 +80,16 @@ module RubyLLM
   end
 end
 
-RubyLLM::Provider.register :openai, RubyLLM::Providers::OpenAI
 RubyLLM::Provider.register :anthropic, RubyLLM::Providers::Anthropic
-RubyLLM::Provider.register :gemini, RubyLLM::Providers::Gemini
-RubyLLM::Provider.register :deepseek, RubyLLM::Providers::DeepSeek
 RubyLLM::Provider.register :bedrock, RubyLLM::Providers::Bedrock
-RubyLLM::Provider.register :openrouter, RubyLLM::Providers::OpenRouter
+RubyLLM::Provider.register :deepseek, RubyLLM::Providers::DeepSeek
+RubyLLM::Provider.register :gemini, RubyLLM::Providers::Gemini
+RubyLLM::Provider.register :gpustack, RubyLLM::Providers::GPUStack
+RubyLLM::Provider.register :mistral, RubyLLM::Providers::Mistral
 RubyLLM::Provider.register :ollama, RubyLLM::Providers::Ollama
+RubyLLM::Provider.register :openai, RubyLLM::Providers::OpenAI
+RubyLLM::Provider.register :openrouter, RubyLLM::Providers::OpenRouter
+RubyLLM::Provider.register :perplexity, RubyLLM::Providers::Perplexity
 
 if defined?(Rails::Railtie)
   require 'ruby_llm/railtie'

data/lib/tasks/models_docs.rake

@@ -6,14 +6,14 @@ require 'fileutils'
 namespace :models do
   desc 'Generate available models documentation'
   task :docs do
-    FileUtils.mkdir_p('docs/guides') # ensure output directory exists
+    FileUtils.mkdir_p('docs') # ensure output directory exists
 
     # Generate markdown content
     output = generate_models_markdown
 
     # Write the output
-    File.write('docs/guides/available-models.md', output)
-    puts 'Generated docs/guides/available-models.md'
+    File.write('docs/available-models.md', output)
+    puts 'Generated docs/available-models.md'
   end
 end
 
@@ -22,15 +22,15 @@ def generate_models_markdown
     ---
     layout: default
     title: Available Models
-    parent: Guides
-    nav_order: 11
-    permalink: /guides/available-models
+    nav_order: 5
+    permalink: /available-models
+    description: Browse hundreds of AI models from every major provider. Always up-to-date, automatically generated.
     ---
 
     # Available Models
     {: .no_toc }
 
-    This guide lists all models available in RubyLLM, automatically generated from the current [model registry](https://github.com/crmne/ruby_llm/blob/main/lib/ruby_llm/models.json).
+    Every model, every provider, always current. Your complete AI model reference.
     {: .fs-6 .fw-300 }
 
     ## Table of contents

data/lib/tasks/models_update.rake

@@ -22,7 +22,9 @@ def configure_from_env
     config.anthropic_api_key = ENV.fetch('ANTHROPIC_API_KEY', nil)
     config.gemini_api_key = ENV.fetch('GEMINI_API_KEY', nil)
     config.deepseek_api_key = ENV.fetch('DEEPSEEK_API_KEY', nil)
+    config.perplexity_api_key = ENV.fetch('PERPLEXITY_API_KEY', nil)
     config.openrouter_api_key = ENV.fetch('OPENROUTER_API_KEY', nil)
+    config.mistral_api_key = ENV.fetch('MISTRAL_API_KEY', nil)
     configure_bedrock(config)
     config.request_timeout = 30
   end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 1.3.2beta1
+  version: 1.5.0
 platform: ruby
 authors:
 - Carmine Paolino
@@ -135,6 +135,15 @@ extra_rdoc_files: []
 files:
 - LICENSE
 - README.md
+- lib/generators/ruby_llm/install/templates/INSTALL_INFO.md.tt
+- lib/generators/ruby_llm/install/templates/chat_model.rb.tt
+- lib/generators/ruby_llm/install/templates/create_chats_migration.rb.tt
+- lib/generators/ruby_llm/install/templates/create_messages_migration.rb.tt
+- lib/generators/ruby_llm/install/templates/create_tool_calls_migration.rb.tt
+- lib/generators/ruby_llm/install/templates/initializer.rb.tt
+- lib/generators/ruby_llm/install/templates/message_model.rb.tt
+- lib/generators/ruby_llm/install/templates/tool_call_model.rb.tt
+- lib/generators/ruby_llm/install_generator.rb
 - lib/ruby_llm.rb
 - lib/ruby_llm/active_record/acts_as.rb
 - lib/ruby_llm/aliases.json
@@ -192,6 +201,14 @@ files:
 - lib/ruby_llm/providers/gemini/models.rb
 - lib/ruby_llm/providers/gemini/streaming.rb
 - lib/ruby_llm/providers/gemini/tools.rb
+- lib/ruby_llm/providers/gpustack.rb
+- lib/ruby_llm/providers/gpustack/chat.rb
+- lib/ruby_llm/providers/gpustack/models.rb
+- lib/ruby_llm/providers/mistral.rb
+- lib/ruby_llm/providers/mistral/capabilities.rb
+- lib/ruby_llm/providers/mistral/chat.rb
+- lib/ruby_llm/providers/mistral/embeddings.rb
+- lib/ruby_llm/providers/mistral/models.rb
 - lib/ruby_llm/providers/ollama.rb
 - lib/ruby_llm/providers/ollama/chat.rb
 - lib/ruby_llm/providers/ollama/media.rb
@@ -206,6 +223,10 @@ files:
 - lib/ruby_llm/providers/openai/tools.rb
 - lib/ruby_llm/providers/openrouter.rb
 - lib/ruby_llm/providers/openrouter/models.rb
+- lib/ruby_llm/providers/perplexity.rb
+- lib/ruby_llm/providers/perplexity/capabilities.rb
+- lib/ruby_llm/providers/perplexity/chat.rb
+- lib/ruby_llm/providers/perplexity/models.rb
 - lib/ruby_llm/railtie.rb
 - lib/ruby_llm/stream_accumulator.rb
 - lib/ruby_llm/streaming.rb