ruby_llm 1.6.1 → 1.6.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e1fd977cc46769d8d235fe13a31e030a133618a335c8cd012ab8de2b7d439297
-  data.tar.gz: ce8d22464541ac19246e6018fde5c7b414ba0022bc5995f8900052f3d0ac5b42
+  metadata.gz: 1b210b6577ab40b05d222354a9cb4c9dd115d7237e4b88be3cd63ee42aa2e122
+  data.tar.gz: 58d0df110e469fef90f5c5591bed4dd3a2d4369afa982217793c11fecb00e320
 SHA512:
-  metadata.gz: aec038a65867c5fa13b93a11aa5356183b8f105bb7f48792e6300a9259a809206bc07da0b54c63a0a0de726cc3698ec2cd79c511be43e6fd8f47273adecc47a9
-  data.tar.gz: 5386da468b9f413d7fbf27192a693c2427a1e88a9082a7375f5641f01854a7cbd2235c1c8108ea870c97c97ea79c712bdd5d33e2ac09a22e25a74f9510a1599f
+  metadata.gz: 3979e1be85c20e8961eb54bcc1d3cf0e8f49dc10fbb5bbbf57e292a098b4c09b053101473c59aef2186c5583276ea5c909b461ba34ac4f8c8eb0f0a5e69cc76d
+  data.tar.gz: 452f86034599f6d8b2d200c3357137f0ec93087b410c4eef5e5393976414b8a986871fa532c6149fb69b081d66f1a823891ab27bd186fb9a31a1ca232707a13d

data/README.md

@@ -6,7 +6,7 @@
 **One *beautiful* Ruby API for GPT, Claude, Gemini, and more.** Easily build chatbots, AI agents, RAG applications, and content generators. Features chat (text, images, audio, PDFs), image generation, embeddings, tools (function calling), structured output, Rails integration, and streaming. Works with OpenAI, Anthropic, Google Gemini, AWS Bedrock, DeepSeek, Mistral, Ollama (local models), OpenRouter, Perplexity, GPUStack, and any OpenAI-compatible API.
 
 <div class="badge-container">
-  <a href="https://badge.fury.io/rb/ruby_llm"><img src="https://badge.fury.io/rb/ruby_llm.svg?a=3" alt="Gem Version" /></a>
+  <a href="https://badge.fury.io/rb/ruby_llm"><img src="https://badge.fury.io/rb/ruby_llm.svg?a=4" alt="Gem Version" /></a>
   <a href="https://github.com/testdouble/standard"><img src="https://img.shields.io/badge/code_style-standard-brightgreen.svg" alt="Ruby Style Guide" /></a>
   <a href="https://rubygems.org/gems/ruby_llm"><img alt="Gem Downloads" src="https://img.shields.io/gem/dt/ruby_llm"></a>
   <a href="https://codecov.io/gh/crmne/ruby_llm"><img src="https://codecov.io/gh/crmne/ruby_llm/branch/main/graph/badge.svg" alt="codecov" /></a>

data/lib/ruby_llm/active_record/acts_as.rb

@@ -234,7 +234,7 @@ module RubyLLM
       end
 
       def persist_new_message
-        @message = messages.create!(role: :assistant, content: String.new)
+        @message = messages.create!(role: :assistant, content: '')
       end
 
       def persist_message_completion(message)

data/lib/ruby_llm/chat.rb

@@ -51,10 +51,6 @@ module RubyLLM
     end
 
     def with_tool(tool)
-      unless @model.supports_functions?
-        raise UnsupportedFunctionsError, "Model #{@model.id} doesn't support function calling"
-      end
-
       tool_instance = tool.is_a?(Class) ? tool.new : tool
       @tools[tool_instance.name.to_sym] = tool_instance
       self
@@ -94,11 +90,7 @@ module RubyLLM
       self
     end
 
-    def with_schema(schema, force: false)
-      unless force || @model.structured_output?
-        raise UnsupportedStructuredOutputError, "Model #{@model.id} doesn't support structured output"
-      end
-
+    def with_schema(schema)
       schema_instance = schema.is_a?(Class) ? schema.new : schema
 
       # Accept both RubyLLM::Schema instances and plain JSON schemas

data/lib/ruby_llm/configuration.rb

@@ -44,7 +44,6 @@ module RubyLLM
                   :logger,
                   :log_file,
                   :log_level,
-                  :log_assume_model_exists,
                   :log_stream_debug
 
     def initialize
@@ -64,7 +63,6 @@ module RubyLLM
       # Logging configuration
       @log_file = $stdout
       @log_level = ENV['RUBYLLM_DEBUG'] ? Logger::DEBUG : Logger::INFO
-      @log_assume_model_exists = true
       @log_stream_debug = ENV['RUBYLLM_STREAM_DEBUG'] == 'true'
     end
 

data/lib/ruby_llm/content.rb

@@ -43,7 +43,7 @@ module RubyLLM
     def process_attachments(attachments)
       if attachments.is_a?(Hash)
         # Ignores types (like :image, :audio, :text, :pdf) since we have robust MIME type detection
-        attachments.each_value(&method(:process_attachments_array_or_string))
+        attachments.each_value { |attachment| process_attachments_array_or_string(attachment) }
       else
         process_attachments_array_or_string attachments
       end

data/lib/ruby_llm/error.rb

@@ -23,9 +23,7 @@ module RubyLLM
   class ConfigurationError < StandardError; end
   class InvalidRoleError < StandardError; end
   class ModelNotFoundError < StandardError; end
-  class UnsupportedFunctionsError < StandardError; end
   class UnsupportedAttachmentError < StandardError; end
-  class UnsupportedStructuredOutputError < StandardError; end
 
   # Error classes for different HTTP status codes
   class BadRequestError < Error; end

data/lib/ruby_llm/models.json

@@ -16285,7 +16285,6 @@
       "supported_parameters": [
         "frequency_penalty",
         "logit_bias",
-        "logprobs",
         "max_tokens",
         "min_p",
         "presence_penalty",
@@ -16297,7 +16296,6 @@
         "tool_choice",
         "tools",
         "top_k",
-        "top_logprobs",
         "top_p"
       ]
     }
@@ -17474,7 +17472,6 @@
       "supported_parameters": [
         "frequency_penalty",
         "logit_bias",
-        "logprobs",
         "max_tokens",
         "min_p",
         "presence_penalty",
@@ -17484,7 +17481,6 @@
         "stop",
         "temperature",
         "top_k",
-        "top_logprobs",
         "top_p"
       ]
     }
@@ -19472,6 +19468,72 @@
       ]
     }
   },
+  {
+    "id": "mistralai/mistral-medium-3.1",
+    "name": "Mistral: Mistral Medium 3.1",
+    "provider": "openrouter",
+    "family": "mistralai",
+    "created_at": "2025-08-13 16:33:59 +0200",
+    "context_window": 131072,
+    "max_output_tokens": null,
+    "knowledge_cutoff": null,
+    "modalities": {
+      "input": [
+        "text",
+        "image"
+      ],
+      "output": [
+        "text"
+      ]
+    },
+    "capabilities": [
+      "streaming",
+      "function_calling",
+      "structured_output"
+    ],
+    "pricing": {
+      "text_tokens": {
+        "standard": {
+          "input_per_million": 0.39999999999999997,
+          "output_per_million": 2.0
+        }
+      }
+    },
+    "metadata": {
+      "description": "Mistral Medium 3.1 is an updated version of Mistral Medium 3, which is a high-performance enterprise-grade language model designed to deliver frontier-level capabilities at significantly reduced operational cost. It balances state-of-the-art reasoning and multimodal performance with 8× lower cost compared to traditional large models, making it suitable for scalable deployments across professional and industrial use cases.\n\nThe model excels in domains such as coding, STEM reasoning, and enterprise adaptation. It supports hybrid, on-prem, and in-VPC deployments and is optimized for integration into custom workflows. Mistral Medium 3.1 offers competitive accuracy relative to larger models like Claude Sonnet 3.5/3.7, Llama 4 Maverick, and Command R+, while maintaining broad compatibility across cloud environments.",
+      "architecture": {
+        "modality": "text+image->text",
+        "input_modalities": [
+          "text",
+          "image"
+        ],
+        "output_modalities": [
+          "text"
+        ],
+        "tokenizer": "Mistral",
+        "instruct_type": null
+      },
+      "top_provider": {
+        "context_length": 131072,
+        "max_completion_tokens": null,
+        "is_moderated": false
+      },
+      "per_request_limits": null,
+      "supported_parameters": [
+        "frequency_penalty",
+        "max_tokens",
+        "presence_penalty",
+        "response_format",
+        "seed",
+        "stop",
+        "structured_outputs",
+        "temperature",
+        "tool_choice",
+        "tools",
+        "top_p"
+      ]
+    }
+  },
   {
     "id": "mistralai/mistral-nemo",
     "name": "Mistral: Mistral Nemo",
@@ -29554,6 +29616,7 @@
         "response_format",
         "seed",
         "stop",
+        "structured_outputs",
         "temperature",
         "tool_choice",
         "tools",
@@ -29700,7 +29763,7 @@
     "name": "Sonar",
     "provider": "perplexity",
     "family": "sonar",
-    "created_at": "2025-08-13 11:35:41 +0200",
+    "created_at": "2025-08-14 00:27:27 +0200",
     "context_window": 128000,
     "max_output_tokens": 4096,
     "knowledge_cutoff": null,
@@ -29732,7 +29795,7 @@
     "name": "Sonar Deep Research",
     "provider": "perplexity",
     "family": "sonar_deep_research",
-    "created_at": "2025-08-13 11:35:41 +0200",
+    "created_at": "2025-08-14 00:27:27 +0200",
     "context_window": 128000,
     "max_output_tokens": 4096,
     "knowledge_cutoff": null,
@@ -29767,7 +29830,7 @@
     "name": "Sonar Pro",
     "provider": "perplexity",
     "family": "sonar_pro",
-    "created_at": "2025-08-13 11:35:41 +0200",
+    "created_at": "2025-08-14 00:27:27 +0200",
     "context_window": 200000,
     "max_output_tokens": 8192,
     "knowledge_cutoff": null,
@@ -29799,7 +29862,7 @@
     "name": "Sonar Reasoning",
     "provider": "perplexity",
     "family": "sonar_reasoning",
-    "created_at": "2025-08-13 11:35:41 +0200",
+    "created_at": "2025-08-14 00:27:27 +0200",
     "context_window": 128000,
     "max_output_tokens": 4096,
     "knowledge_cutoff": null,
@@ -29831,7 +29894,7 @@
     "name": "Sonar Reasoning Pro",
     "provider": "perplexity",
     "family": "sonar_reasoning_pro",
-    "created_at": "2025-08-13 11:35:41 +0200",
+    "created_at": "2025-08-14 00:27:27 +0200",
     "context_window": 128000,
     "max_output_tokens": 8192,
     "knowledge_cutoff": null,

data/lib/ruby_llm/models.rb

@@ -64,16 +64,12 @@ module RubyLLM
 
           model = Model::Info.new(
             id: model_id,
-            name: model_id.gsub('-', ' ').capitalize,
+            name: model_id.tr('-', ' ').capitalize,
             provider: provider_instance.slug,
             capabilities: %w[function_calling streaming],
             modalities: { input: %w[text image], output: %w[text] },
             metadata: { warning: 'Assuming model exists, capabilities may not be accurate' }
           )
-          if RubyLLM.config.log_assume_model_exists
-            RubyLLM.logger.warn "Assuming model '#{model_id}' exists for provider '#{provider}'. " \
-                                'Capabilities may not be accurately reflected.'
-          end
         else
           model = Models.find model_id, provider
           provider_class = Provider.providers[model.provider.to_sym] || raise(Error,

data/lib/ruby_llm/providers/bedrock/streaming/base.rb

@@ -47,7 +47,7 @@ module RubyLLM
           end
 
           def handle_stream(&block)
-            buffer = String.new
+            buffer = +''
             proc do |chunk, _bytes, env|
               if env && env.status != 200
                 handle_failed_response(chunk, buffer, env)

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

@@ -80,7 +80,7 @@ module RubyLLM
             content: extract_content(data),
             tool_calls: tool_calls,
             input_tokens: data.dig('usageMetadata', 'promptTokenCount'),
-            output_tokens: data.dig('usageMetadata', 'candidatesTokenCount'),
+            output_tokens: calculate_output_tokens(data),
             model_id: data['modelVersion'] || response.env.url.path.split('/')[3].split(':')[0],
             raw: response
           )
@@ -133,6 +133,12 @@ module RubyLLM
           parts = candidate.dig('content', 'parts')
           parts&.any? { |p| p['functionCall'] }
         end
+
+        def calculate_output_tokens(data)
+          candidates = data.dig('usageMetadata', 'candidatesTokenCount') || 0
+          thoughts = data.dig('usageMetadata', 'thoughtsTokenCount') || 0
+          candidates + thoughts
+        end
       end
     end
   end

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

@@ -42,7 +42,10 @@ module RubyLLM
         end
 
         def extract_output_tokens(data)
-          data.dig('usageMetadata', 'candidatesTokenCount')
+          candidates = data.dig('usageMetadata', 'candidatesTokenCount') || 0
+          thoughts = data.dig('usageMetadata', 'thoughtsTokenCount') || 0
+          total = candidates + thoughts
+          total.positive? ? total : nil
         end
 
         def parse_streaming_error(data)

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

@@ -198,11 +198,11 @@ module RubyLLM
             .gsub(/(\d{4}) (\d{2}) (\d{2})/, '\1\2\3')
             .gsub(/^(?:Gpt|Chatgpt|Tts|Dall E) /) { |m| special_prefix_format(m.strip) }
             .gsub(/^O([13]) /, 'O\1-')
-            .gsub(/^O[13] Mini/, '\0'.gsub(' ', '-'))
+            .gsub(/^O[13] Mini/, '\0'.tr(' ', '-'))
             .gsub(/\d\.\d /, '\0'.sub(' ', '-'))
             .gsub(/4o (?=Mini|Preview|Turbo|Audio|Realtime|Transcribe|Tts)/, '4o-')
             .gsub(/\bHd\b/, 'HD')
-            .gsub(/(?:Omni|Text) Moderation/, '\0'.gsub(' ', '-'))
+            .gsub(/(?:Omni|Text) Moderation/, '\0'.tr(' ', '-'))
             .gsub('Text Embedding', 'text-embedding-')
         end
 

data/lib/ruby_llm/stream_accumulator.rb

@@ -8,7 +8,7 @@ module RubyLLM
     attr_reader :content, :model_id, :tool_calls
 
     def initialize
-      @content = String.new
+      @content = +''
       @tool_calls = {}
       @input_tokens = 0
       @output_tokens = 0
@@ -66,7 +66,7 @@ module RubyLLM
       new_tool_calls.each_value do |tool_call|
         if tool_call.id
           tool_call_id = tool_call.id.empty? ? SecureRandom.uuid : tool_call.id
-          tool_call_arguments = tool_call.arguments.empty? ? String.new : tool_call.arguments
+          tool_call_arguments = tool_call.arguments.empty? ? +'' : tool_call.arguments
           @tool_calls[tool_call.id] = ToolCall.new(
             id: tool_call_id,
             name: tool_call.name,

data/lib/ruby_llm/streaming.rb

@@ -43,7 +43,7 @@ module RubyLLM
     private
 
     def to_json_stream(&)
-      buffer = String.new
+      buffer = +''
       parser = EventStreamParser::Parser.new
 
       create_stream_processor(parser, buffer, &)

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '1.6.1'
+  VERSION = '1.6.2'
 end

data/lib/tasks/aliases.rake

@@ -65,7 +65,7 @@ namespace :aliases do # rubocop:disable Metrics/BlockLength
 
       base_name = Regexp.last_match(1)
       # Normalize to Anthropic naming convention
-      anthropic_name = base_name.gsub('.', '-')
+      anthropic_name = base_name.tr('.', '-')
 
       # Skip if we already have an alias for this
       next if aliases[anthropic_name]
@@ -91,7 +91,7 @@ namespace :aliases do # rubocop:disable Metrics/BlockLength
       # OpenRouter uses "google/" prefix and sometimes different naming
       openrouter_variants = [
         "google/#{model}",
-        "google/#{model.gsub('gemini-', 'gemini-').gsub('.', '-')}",
+        "google/#{model.gsub('gemini-', 'gemini-').tr('.', '-')}",
         "google/#{model.gsub('gemini-', 'gemini-')}"
       ]
 

data/lib/tasks/models_docs.rake

@@ -86,7 +86,7 @@ def generate_models_markdown
 end
 
 def generate_provider_sections
-  RubyLLM::Provider.providers.map do |provider, provider_class|
+  RubyLLM::Provider.providers.filter_map do |provider, provider_class|
     models = RubyLLM.models.by_provider(provider)
     next if models.none?
 
@@ -95,7 +95,7 @@ def generate_provider_sections
 
       #{models_table(models)}
     PROVIDER
-  end.compact.join("\n\n")
+  end.join("\n\n")
 end
 
 def generate_capability_sections
@@ -107,7 +107,7 @@ def generate_capability_sections
     'Batch Processing' => RubyLLM.models.select { |m| m.capabilities.include?('batch') }
   }
 
-  capabilities.map do |capability, models|
+  capabilities.filter_map do |capability, models|
     next if models.none?
 
     <<~CAPABILITY
@@ -115,7 +115,7 @@ def generate_capability_sections
 
       #{models_table(models)}
     CAPABILITY
-  end.compact.join("\n\n")
+  end.join("\n\n")
 end
 
 def generate_modality_sections # rubocop:disable Metrics/PerceivedComplexity

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 1.6.1
+  version: 1.6.2
 platform: ruby
 authors:
 - Carmine Paolino