ruby_llm 1.1.0 → 1.1.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b157007e50a6d43f11591847306e9e950904ad71dde849dded2f4364f376fa0f
-  data.tar.gz: 9f2024f254134590b971a98b6fe1cae161bfd25378f747cd6c192237add0d80c
+  metadata.gz: d3226fdaf84de31b1deefae4aa85ca721d4fc6ce2a84572d211a57a0acbf1cb3
+  data.tar.gz: 8846beaefacc53539987eaf6b619e4833267a928f688ad6afc4277180eaac7b2
 SHA512:
-  metadata.gz: d804afb295a9b9d174f44ac110ce60e9d6b3ab6a4e96d003532bbdf5c95bc6db34b7affcec44d6025eff98d71a38a356417dee3912776b44feb735b803c49ed6
-  data.tar.gz: ef562bad49590fc86fb78311f861688175a55b9ab2b6d68c23d74cbaeabc1ecf9703b420bed5f8bc689fb1542cbe2fe484b1aa52e8df53d8b6fedb6737b19f18
+  metadata.gz: 1556e816f7ffdb9323b53067f6faefc9c96eb0e116740b690dfa96afdc39c96d0b294efe4b8a363279bf6d43c907d4e6ebdf92cde0dc996218961627a655a626
+  data.tar.gz: 75def96179be6d1d44cce576a698aaeb7bf5cf11ac60d0699f12de7d7dfa32f9a71906821ec43d1f77b98fd152156af41d10b89e832406601f42239872950a8c

data/lib/ruby_llm/active_record/acts_as.rb

@@ -25,13 +25,13 @@ module RubyLLM
                    to: :to_llm
         end
 
-        def acts_as_message(chat_class: 'Chat', tool_call_class: 'ToolCall') # rubocop:disable Metrics/MethodLength
+        def acts_as_message(chat_class: 'Chat', tool_call_class: 'ToolCall', touch_chat: false) # rubocop:disable Metrics/MethodLength
           include MessageMethods
 
           @chat_class = chat_class.to_s
           @tool_call_class = tool_call_class.to_s
 
-          belongs_to :chat, class_name: @chat_class
+          belongs_to :chat, class_name: @chat_class, touch: touch_chat
           has_many :tool_calls, class_name: @tool_call_class, dependent: :destroy
 
           belongs_to :parent_tool_call,

data/lib/ruby_llm/aliases.json

@@ -34,32 +34,5 @@
   "claude-2-1": {
     "anthropic": "claude-2.1",
     "bedrock": "anthropic.claude-2.1"
-  },
-  "gpt-4o": {
-    "openai": "gpt-4o-2024-11-20"
-  },
-  "gpt-4o-mini": {
-    "openai": "gpt-4o-mini-2024-07-18"
-  },
-  "gpt-4-turbo": {
-    "openai": "gpt-4-turbo-2024-04-09"
-  },
-  "gemini-1.5-flash": {
-    "gemini": "gemini-1.5-flash-002"
-  },
-  "gemini-1.5-flash-8b": {
-    "gemini": "gemini-1.5-flash-8b-001"
-  },
-  "gemini-1.5-pro": {
-    "gemini": "gemini-1.5-pro-002"
-  },
-  "gemini-2.0-flash": {
-    "gemini": "gemini-2.0-flash-001"
-  },
-  "o1": {
-    "openai": "o1-2024-12-17"
-  },
-  "o3-mini": {
-    "openai": "o3-mini-2025-01-31"
   }
 }

data/lib/ruby_llm/models.json

@@ -4,7 +4,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Haiku",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_5_haiku",
@@ -34,7 +34,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -47,7 +47,8 @@
       "provider_name": "Anthropic",
       "customizations_supported": [],
       "inference_configurations": [
-        "ON_DEMAND"
+        "ON_DEMAND",
+        "INFERENCE_PROFILE"
       ],
       "response_streaming_supported": true,
       "input_modalities": [
@@ -64,7 +65,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -94,7 +95,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -124,7 +125,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -154,7 +155,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet v2",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -184,7 +185,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet v2",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -214,7 +215,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet v2",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -244,7 +245,7 @@
     "created_at": null,
     "display_name": "Claude 3.5 Sonnet v2",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -270,11 +271,11 @@
     }
   },
   {
-    "id": "us.anthropic.claude-3-7-sonnet-20250219-v1:0",
+    "id": "anthropic.claude-3-7-sonnet-20250219-v1:0",
     "created_at": null,
     "display_name": "Claude 3.7 Sonnet",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude3_sonnet",
@@ -607,13 +608,13 @@
     "created_at": null,
     "display_name": "Claude Instant",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude_instant",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 0.8,
     "output_price_per_million": 2.4,
     "metadata": {
@@ -636,13 +637,13 @@
     "created_at": null,
     "display_name": "Claude Instant",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude_instant",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 0.8,
     "output_price_per_million": 2.4,
     "metadata": {
@@ -665,13 +666,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -694,13 +695,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -723,13 +724,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -752,13 +753,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -781,13 +782,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -810,13 +811,13 @@
     "created_at": null,
     "display_name": "Claude",
     "provider": "bedrock",
-    "context_window": 4096,
+    "context_window": 200000,
     "max_tokens": 4096,
     "type": "chat",
     "family": "claude2",
-    "supports_vision": false,
-    "supports_functions": false,
-    "supports_json_mode": false,
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
     "input_price_per_million": 8.0,
     "output_price_per_million": 24.0,
     "metadata": {
@@ -1613,7 +1614,8 @@
       "output_token_limit": 8192,
       "supported_generation_methods": [
         "generateContent",
-        "countTokens"
+        "countTokens",
+        "createCachedContent"
       ]
     }
   },
@@ -1638,7 +1640,8 @@
       "output_token_limit": 8192,
       "supported_generation_methods": [
         "generateContent",
-        "countTokens"
+        "countTokens",
+        "createCachedContent"
       ]
     }
   },
@@ -1794,6 +1797,31 @@
       ]
     }
   },
+  {
+    "id": "gemini-2.0-flash-live-001",
+    "created_at": null,
+    "display_name": "Gemini 2.0 Flash 001",
+    "provider": "gemini",
+    "context_window": 131072,
+    "max_tokens": 8192,
+    "type": "chat",
+    "family": "gemini20_flash",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.1,
+    "output_price_per_million": 0.4,
+    "metadata": {
+      "version": "001",
+      "description": "Gemini 2.0 Flash 001",
+      "input_token_limit": 131072,
+      "output_token_limit": 8192,
+      "supported_generation_methods": [
+        "bidiGenerateContent",
+        "countTokens"
+      ]
+    }
+  },
   {
     "id": "gemini-2.0-flash-thinking-exp",
     "created_at": null,
@@ -1944,6 +1972,31 @@
       ]
     }
   },
+  {
+    "id": "gemini-2.5-pro-preview-03-25",
+    "created_at": null,
+    "display_name": "Gemini 2.5 Pro Preview 03-25",
+    "provider": "gemini",
+    "context_window": 1048576,
+    "max_tokens": 65536,
+    "type": "chat",
+    "family": "other",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.075,
+    "output_price_per_million": 0.3,
+    "metadata": {
+      "version": "2.5-preview-03-25",
+      "description": "Gemini 2.5 Pro Preview 03-25",
+      "input_token_limit": 1048576,
+      "output_token_limit": 65536,
+      "supported_generation_methods": [
+        "generateContent",
+        "countTokens"
+      ]
+    }
+  },
   {
     "id": "gemini-embedding-exp",
     "created_at": null,
@@ -1964,7 +2017,8 @@
       "input_token_limit": 8192,
       "output_token_limit": 1,
       "supported_generation_methods": [
-        "embedContent"
+        "embedContent",
+        "countTextTokens"
       ]
     }
   },
@@ -1988,7 +2042,8 @@
       "input_token_limit": 8192,
       "output_token_limit": 1,
       "supported_generation_methods": [
-        "embedContent"
+        "embedContent",
+        "countTextTokens"
       ]
     }
   },
@@ -2042,6 +2097,56 @@
       ]
     }
   },
+  {
+    "id": "gemma-3-12b-it",
+    "created_at": null,
+    "display_name": "Gemma 3 12B",
+    "provider": "gemini",
+    "context_window": 32768,
+    "max_tokens": 8192,
+    "type": "chat",
+    "family": "other",
+    "supports_vision": false,
+    "supports_functions": false,
+    "supports_json_mode": false,
+    "input_price_per_million": 0.075,
+    "output_price_per_million": 0.3,
+    "metadata": {
+      "version": "001",
+      "description": null,
+      "input_token_limit": 32768,
+      "output_token_limit": 8192,
+      "supported_generation_methods": [
+        "generateContent",
+        "countTokens"
+      ]
+    }
+  },
+  {
+    "id": "gemma-3-1b-it",
+    "created_at": null,
+    "display_name": "Gemma 3 1B",
+    "provider": "gemini",
+    "context_window": 32768,
+    "max_tokens": 8192,
+    "type": "chat",
+    "family": "other",
+    "supports_vision": false,
+    "supports_functions": false,
+    "supports_json_mode": false,
+    "input_price_per_million": 0.075,
+    "output_price_per_million": 0.3,
+    "metadata": {
+      "version": "001",
+      "description": null,
+      "input_token_limit": 32768,
+      "output_token_limit": 8192,
+      "supported_generation_methods": [
+        "generateContent",
+        "countTokens"
+      ]
+    }
+  },
   {
     "id": "gemma-3-27b-it",
     "created_at": null,
@@ -2067,6 +2172,31 @@
       ]
     }
   },
+  {
+    "id": "gemma-3-4b-it",
+    "created_at": null,
+    "display_name": "Gemma 3 4B",
+    "provider": "gemini",
+    "context_window": 32768,
+    "max_tokens": 8192,
+    "type": "chat",
+    "family": "other",
+    "supports_vision": false,
+    "supports_functions": false,
+    "supports_json_mode": false,
+    "input_price_per_million": 0.075,
+    "output_price_per_million": 0.3,
+    "metadata": {
+      "version": "001",
+      "description": null,
+      "input_token_limit": 32768,
+      "output_token_limit": 8192,
+      "supported_generation_methods": [
+        "generateContent",
+        "countTokens"
+      ]
+    }
+  },
   {
     "id": "gpt-3.5-turbo",
     "created_at": "2023-02-28T19:56:42+01:00",
@@ -2314,6 +2444,120 @@
       "owned_by": "system"
     }
   },
+  {
+    "id": "gpt-4.1",
+    "created_at": "2025-04-10T22:22:22+02:00",
+    "display_name": "GPT-4.1",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 2.0,
+    "output_price_per_million": 8.0,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
+  {
+    "id": "gpt-4.1-2025-04-14",
+    "created_at": "2025-04-10T22:09:06+02:00",
+    "display_name": "GPT-4.1 20250414",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 2.0,
+    "output_price_per_million": 8.0,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
+  {
+    "id": "gpt-4.1-mini",
+    "created_at": "2025-04-10T22:49:33+02:00",
+    "display_name": "GPT-4.1 Mini",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41_mini",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.4,
+    "output_price_per_million": 1.6,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
+  {
+    "id": "gpt-4.1-mini-2025-04-14",
+    "created_at": "2025-04-10T22:39:07+02:00",
+    "display_name": "GPT-4.1 Mini 20250414",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41_mini",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.4,
+    "output_price_per_million": 1.6,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
+  {
+    "id": "gpt-4.1-nano",
+    "created_at": "2025-04-10T23:48:27+02:00",
+    "display_name": "GPT-4.1 Nano",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41_nano",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.1,
+    "output_price_per_million": 0.4,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
+  {
+    "id": "gpt-4.1-nano-2025-04-14",
+    "created_at": "2025-04-10T23:37:05+02:00",
+    "display_name": "GPT-4.1 Nano 20250414",
+    "provider": "openai",
+    "context_window": 1047576,
+    "max_tokens": 32768,
+    "type": "chat",
+    "family": "gpt41_nano",
+    "supports_vision": true,
+    "supports_functions": true,
+    "supports_json_mode": true,
+    "input_price_per_million": 0.1,
+    "output_price_per_million": 0.4,
+    "metadata": {
+      "object": "model",
+      "owned_by": "system"
+    }
+  },
   {
     "id": "gpt-4.5-preview",
     "created_at": "2025-02-27T03:24:19+01:00",
@@ -3249,6 +3493,30 @@
       "owned_by": "system"
     }
   },
+  {
+    "id": "veo-2.0-generate-001",
+    "created_at": null,
+    "display_name": "Veo 2",
+    "provider": "gemini",
+    "context_window": 480,
+    "max_tokens": 8192,
+    "type": "chat",
+    "family": "other",
+    "supports_vision": false,
+    "supports_functions": false,
+    "supports_json_mode": false,
+    "input_price_per_million": 0.075,
+    "output_price_per_million": 0.3,
+    "metadata": {
+      "version": "2.0",
+      "description": "Vertex served Veo 2 model.",
+      "input_token_limit": 480,
+      "output_token_limit": 8192,
+      "supported_generation_methods": [
+        "predictLongRunning"
+      ]
+    }
+  },
   {
     "id": "whisper-1",
     "created_at": "2023-02-27T22:13:04+01:00",

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

@@ -3,13 +3,15 @@
 module RubyLLM
   module Providers
     module OpenAI
-      # Determines capabilities and pricing for OpenAI models
-      module Capabilities # rubocop:disable Metrics/ModuleLength
+      module Capabilities # rubocop:disable Metrics/ModuleLength,Style/Documentation
         module_function
 
         MODEL_PATTERNS = {
           dall_e: /^dall-e/,
           chatgpt4o: /^chatgpt-4o/,
+          gpt41: /^gpt-4\.1(?!-(?:mini|nano))/,
+          gpt41_mini: /^gpt-4\.1-mini/,
+          gpt41_nano: /^gpt-4\.1-nano/,
           gpt4: /^gpt-4(?:-\d{6})?$/,
           gpt4_turbo: /^gpt-4(?:\.5)?-(?:\d{6}-)?(preview|turbo)/,
           gpt35_turbo: /^gpt-3\.5-turbo/,
@@ -38,8 +40,9 @@ module RubyLLM
           moderation: /^(?:omni|text)-moderation/
         }.freeze
 
-        def context_window_for(model_id) # rubocop:disable Metrics/MethodLength
+        def context_window_for(model_id) # rubocop:disable Metrics/CyclomaticComplexity,Metrics/MethodLength
           case model_family(model_id)
+          when 'gpt41', 'gpt41_mini', 'gpt41_nano' then 1_047_576
           when 'chatgpt4o', 'gpt4_turbo', 'gpt4o', 'gpt4o_audio', 'gpt4o_mini',
                'gpt4o_mini_audio', 'gpt4o_mini_realtime', 'gpt4o_realtime',
                'gpt4o_search', 'gpt4o_transcribe', 'gpt4o_mini_search', 'o1_mini' then 128_000
@@ -55,6 +58,7 @@ module RubyLLM
 
         def max_tokens_for(model_id) # rubocop:disable Metrics/CyclomaticComplexity,Metrics/MethodLength
           case model_family(model_id)
+          when 'gpt41', 'gpt41_mini', 'gpt41_nano' then 32_768
           when 'chatgpt4o', 'gpt4o', 'gpt4o_mini', 'gpt4o_mini_search' then 16_384
           when 'babbage', 'davinci' then 16_384 # rubocop:disable Lint/DuplicateBranch
           when 'gpt4' then 8_192
@@ -71,15 +75,16 @@ module RubyLLM
 
         def supports_vision?(model_id)
           case model_family(model_id)
-          when 'chatgpt4o', 'gpt4', 'gpt4_turbo', 'gpt4o', 'gpt4o_mini', 'o1', 'o1_pro',
-               'moderation', 'gpt4o_search', 'gpt4o_mini_search' then true
+          when 'gpt41', 'gpt41_mini', 'gpt41_nano', 'chatgpt4o', 'gpt4', 'gpt4_turbo', 'gpt4o', 'gpt4o_mini', 'o1',
+               'o1_pro', 'moderation', 'gpt4o_search', 'gpt4o_mini_search' then true
           else false
           end
         end
 
         def supports_functions?(model_id)
           case model_family(model_id)
-          when 'gpt4', 'gpt4_turbo', 'gpt4o', 'gpt4o_mini', 'o1', 'o1_pro', 'o3_mini' then true
+          when 'gpt41', 'gpt41_mini', 'gpt41_nano', 'gpt4', 'gpt4_turbo', 'gpt4o', 'gpt4o_mini', 'o1', 'o1_pro',
+               'o3_mini' then true
           when 'chatgpt4o', 'gpt35_turbo', 'o1_mini', 'gpt4o_mini_tts',
                'gpt4o_transcribe', 'gpt4o_search', 'gpt4o_mini_search' then false
           else false # rubocop:disable Lint/DuplicateBranch
@@ -88,7 +93,8 @@ module RubyLLM
 
         def supports_structured_output?(model_id)
           case model_family(model_id)
-          when 'chatgpt4o', 'gpt4o', 'gpt4o_mini', 'o1', 'o1_pro', 'o3_mini' then true
+          when 'gpt41', 'gpt41_mini', 'gpt41_nano', 'chatgpt4o', 'gpt4o', 'gpt4o_mini', 'o1', 'o1_pro',
+               'o3_mini' then true
           else false
           end
         end
@@ -98,6 +104,9 @@ module RubyLLM
         end
 
         PRICES = {
+          gpt41: { input: 2.0, output: 8.0, cached_input: 0.5 },
+          gpt41_mini: { input: 0.4, output: 1.6, cached_input: 0.1 },
+          gpt41_nano: { input: 0.1, output: 0.4 },
           chatgpt4o: { input: 5.0, output: 15.0 },
           gpt4: { input: 10.0, output: 30.0 },
           gpt4_turbo: { input: 10.0, output: 30.0 },
@@ -141,6 +150,12 @@ module RubyLLM
           prices[:input] || prices[:price] || default_input_price
         end
 
+        def cached_input_price_for(model_id)
+          family = model_family(model_id).to_sym
+          prices = PRICES.fetch(family, {})
+          prices[:cached_input]
+        end
+
         def output_price_for(model_id)
           family = model_family(model_id).to_sym
           prices = PRICES.fetch(family, { output: default_output_price })

data/lib/ruby_llm/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module RubyLLM
-  VERSION = '1.1.0'
+  VERSION = '1.1.2'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: ruby_llm
 version: !ruby/object:Gem::Version
-  version: 1.1.0
+  version: 1.1.2
 platform: ruby
 authors:
 - Carmine Paolino
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-04-08 00:00:00.000000000 Z
+date: 2025-04-15 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: base64
@@ -66,6 +66,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '1'
+- !ruby/object:Gem::Dependency
+  name: faraday-net_http
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '3'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '3'
 - !ruby/object:Gem::Dependency
   name: faraday-retry
   requirement: !ruby/object:Gem::Requirement