llmist 8.1.4 → 9.0.0

package/dist/index.cjs

@@ -277,12 +277,20 @@ var init_model_shortcuts = __esm({
   "src/core/model-shortcuts.ts"() {
     "use strict";
     MODEL_ALIASES = {
-      // OpenAI aliases
+      // OpenAI aliases - GPT-5.2 is the latest flagship
       gpt4: "openai:gpt-4o",
       gpt4o: "openai:gpt-4o",
-      gpt5: "openai:gpt-5",
+      "gpt4o-mini": "openai:gpt-4o-mini",
+      gpt5: "openai:gpt-5.2",
+      // Latest flagship
+      "gpt5.2": "openai:gpt-5.2",
+      "gpt5.1": "openai:gpt-5.1",
       "gpt5-mini": "openai:gpt-5-mini",
       "gpt5-nano": "openai:gpt-5-nano",
+      "gpt5-codex": "openai:gpt-5-codex",
+      o1: "openai:o1",
+      o3: "openai:o3",
+      "o4-mini": "openai:o4-mini",
       // Anthropic aliases
       sonnet: "anthropic:claude-sonnet-4-5",
       "claude-sonnet": "anthropic:claude-sonnet-4-5",
@@ -290,11 +298,12 @@ var init_model_shortcuts = __esm({
       "claude-haiku": "anthropic:claude-haiku-4-5",
       opus: "anthropic:claude-opus-4-5",
       "claude-opus": "anthropic:claude-opus-4-5",
-      // Gemini aliases
-      flash: "gemini:gemini-2.0-flash",
-      "gemini-flash": "gemini:gemini-2.0-flash",
-      "gemini-pro": "gemini:gemini-2.5-pro",
-      pro: "gemini:gemini-2.5-pro"
+      // Gemini aliases - 2.5 Flash is the recommended fast model
+      flash: "gemini:gemini-2.5-flash",
+      "gemini-flash": "gemini:gemini-2.5-flash",
+      "flash-lite": "gemini:gemini-2.5-flash-lite",
+      "gemini-pro": "gemini:gemini-3-pro-preview",
+      pro: "gemini:gemini-3-pro-preview"
     };
     KNOWN_MODEL_PATTERNS = [
       /^gpt-?\d/i,
@@ -3556,10 +3565,10 @@ var init_anthropic_models = __esm({
         contextWindow: 2e5,
         maxOutputTokens: 64e3,
         pricing: {
-          input: 1,
-          output: 5,
-          cachedInput: 0.1,
-          cacheWriteInput: 1.25
+          input: 0.8,
+          output: 4,
+          cachedInput: 0.08,
+          cacheWriteInput: 1
         },
         knowledgeCutoff: "2025-02",
         features: {
@@ -3721,6 +3730,32 @@ var init_anthropic_models = __esm({
           notes: "Legacy model - upgrade to Haiku 4.5 for better performance"
         }
       },
+      // Dated Opus 4.5
+      {
+        provider: "anthropic",
+        modelId: "claude-opus-4-5-20251124",
+        displayName: "Claude Opus 4.5",
+        contextWindow: 2e5,
+        maxOutputTokens: 64e3,
+        pricing: {
+          input: 5,
+          output: 25,
+          cachedInput: 0.5,
+          cacheWriteInput: 6.25
+        },
+        knowledgeCutoff: "2025-03",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true
+        },
+        metadata: {
+          family: "Claude 4",
+          releaseDate: "2025-11-24",
+          notes: "Most powerful model for coding and computer use. Extended thinking support."
+        }
+      },
       // Modern aliases (recommended by Anthropic)
       {
         provider: "anthropic",
@@ -3729,10 +3764,10 @@ var init_anthropic_models = __esm({
         contextWindow: 2e5,
         maxOutputTokens: 64e3,
         pricing: {
-          input: 1,
-          output: 5,
-          cachedInput: 0.1,
-          cacheWriteInput: 1.25
+          input: 0.8,
+          output: 4,
+          cachedInput: 0.08,
+          cacheWriteInput: 1
         },
         knowledgeCutoff: "2025-02",
         features: {
@@ -4318,6 +4353,33 @@ var init_gemini_models = __esm({
           notes: "Best model for multimodal understanding, agentic and vibe-coding. Deep Think mode available."
         }
       },
+      // Gemini 3 Flash (Preview)
+      {
+        provider: "gemini",
+        modelId: "gemini-3-flash-preview",
+        displayName: "Gemini 3 Flash (Preview)",
+        contextWindow: 1048576,
+        maxOutputTokens: 65536,
+        pricing: {
+          input: 0.4,
+          // $0.40 for text/image/video
+          output: 3,
+          cachedInput: 0.04
+        },
+        knowledgeCutoff: "2025-01",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "Gemini 3",
+          releaseDate: "2025-12",
+          notes: "Fast, cost-effective model with Deep Think mode. Good for agentic tasks."
+        }
+      },
       // Gemini 2.5 Pro
       {
         provider: "gemini",
@@ -5052,13 +5114,37 @@ var GPT_IMAGE_SIZES, GPT_IMAGE_QUALITIES, DALLE3_SIZES, DALLE3_QUALITIES, DALLE2
 var init_openai_image_models = __esm({
   "src/providers/openai-image-models.ts"() {
     "use strict";
-    GPT_IMAGE_SIZES = ["1024x1024", "1024x1536", "1536x1024"];
+    GPT_IMAGE_SIZES = ["1024x1024", "1024x1536", "1536x1024", "1920x1080", "auto"];
     GPT_IMAGE_QUALITIES = ["low", "medium", "high"];
     DALLE3_SIZES = ["1024x1024", "1024x1792", "1792x1024"];
     DALLE3_QUALITIES = ["standard", "hd"];
     DALLE2_SIZES = ["256x256", "512x512", "1024x1024"];
     openaiImageModels = [
-      // GPT Image 1 Family (flagship)
+      // GPT Image 1.5 Family (flagship)
+      {
+        provider: "openai",
+        modelId: "gpt-image-1.5",
+        displayName: "GPT Image 1.5",
+        pricing: {
+          bySize: {
+            "1024x1024": { low: 8e-3, medium: 0.03, high: 0.13 },
+            "1024x1536": { low: 0.012, medium: 0.045, high: 0.195 },
+            "1536x1024": { low: 0.012, medium: 0.045, high: 0.195 },
+            "1920x1080": { low: 0.016, medium: 0.06, high: 0.26 }
+          }
+        },
+        supportedSizes: [...GPT_IMAGE_SIZES],
+        supportedQualities: [...GPT_IMAGE_QUALITIES],
+        maxImages: 1,
+        defaultSize: "1024x1024",
+        defaultQuality: "medium",
+        features: {
+          textRendering: true,
+          transparency: true,
+          editing: true
+        }
+      },
+      // GPT Image 1 Family (previous gen)
       {
         provider: "openai",
         modelId: "gpt-image-1",
@@ -5070,7 +5156,7 @@ var init_openai_image_models = __esm({
             "1536x1024": { low: 0.016, medium: 0.06, high: 0.25 }
           }
         },
-        supportedSizes: [...GPT_IMAGE_SIZES],
+        supportedSizes: ["1024x1024", "1024x1536", "1536x1024"],
         supportedQualities: [...GPT_IMAGE_QUALITIES],
         maxImages: 1,
         defaultSize: "1024x1024",
@@ -5091,7 +5177,7 @@ var init_openai_image_models = __esm({
             "1536x1024": { low: 75e-4, medium: 0.03, high: 0.078 }
           }
         },
-        supportedSizes: [...GPT_IMAGE_SIZES],
+        supportedSizes: ["1024x1024", "1024x1536", "1536x1024"],
         supportedQualities: [...GPT_IMAGE_QUALITIES],
         maxImages: 1,
         defaultSize: "1024x1024",
@@ -5101,11 +5187,53 @@ var init_openai_image_models = __esm({
           transparency: true
         }
       },
-      // DALL-E Family
+      // Sora Video Generation Models
+      {
+        provider: "openai",
+        modelId: "sora-2",
+        displayName: "Sora 2",
+        pricing: {
+          bySize: {
+            "1920x1080": { standard: 0.5, high: 1 },
+            "1080x1920": { standard: 0.5, high: 1 },
+            "1024x1024": { standard: 0.4, high: 0.8 }
+          }
+        },
+        supportedSizes: ["1920x1080", "1080x1920", "1024x1024"],
+        supportedQualities: ["standard", "high"],
+        maxImages: 1,
+        defaultSize: "1920x1080",
+        defaultQuality: "standard",
+        features: {
+          videoGeneration: true
+        }
+      },
+      {
+        provider: "openai",
+        modelId: "sora-2-pro",
+        displayName: "Sora 2 Pro",
+        pricing: {
+          bySize: {
+            "1920x1080": { standard: 1, high: 2 },
+            "1080x1920": { standard: 1, high: 2 },
+            "1024x1024": { standard: 0.8, high: 1.6 }
+          }
+        },
+        supportedSizes: ["1920x1080", "1080x1920", "1024x1024"],
+        supportedQualities: ["standard", "high"],
+        maxImages: 1,
+        defaultSize: "1920x1080",
+        defaultQuality: "standard",
+        features: {
+          videoGeneration: true,
+          extendedDuration: true
+        }
+      },
+      // DALL-E Family (deprecated - use GPT Image models instead)
       {
         provider: "openai",
         modelId: "dall-e-3",
-        displayName: "DALL-E 3",
+        displayName: "DALL-E 3 (Deprecated)",
         pricing: {
           bySize: {
             "1024x1024": { standard: 0.04, hd: 0.08 },
@@ -5126,7 +5254,7 @@ var init_openai_image_models = __esm({
       {
         provider: "openai",
         modelId: "dall-e-2",
-        displayName: "DALL-E 2 (Legacy)",
+        displayName: "DALL-E 2 (Deprecated)",
         pricing: {
           bySize: {
             "256x256": 0.016,
@@ -5148,12 +5276,65 @@ var init_openai_models = __esm({
   "src/providers/openai-models.ts"() {
     "use strict";
     OPENAI_MODELS = [
-      // GPT-5 Family
+      // GPT-5.2 Family (Latest flagship)
+      {
+        provider: "openai",
+        modelId: "gpt-5.2",
+        displayName: "GPT-5.2",
+        contextWindow: 1e6,
+        maxOutputTokens: 128e3,
+        pricing: {
+          input: 1.25,
+          output: 10,
+          cachedInput: 0.125
+        },
+        knowledgeCutoff: "2025-03-31",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true,
+          fineTuning: true
+        },
+        metadata: {
+          family: "GPT-5.2",
+          releaseDate: "2025-12-01",
+          notes: "Latest flagship model with 1M context window and enhanced reasoning.",
+          supportsTemperature: false
+        }
+      },
+      {
+        provider: "openai",
+        modelId: "gpt-5.2-pro",
+        displayName: "GPT-5.2 Pro",
+        contextWindow: 1e6,
+        maxOutputTokens: 128e3,
+        pricing: {
+          input: 15,
+          output: 120
+        },
+        knowledgeCutoff: "2025-03-31",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "GPT-5.2",
+          releaseDate: "2025-12-01",
+          notes: "Premium tier GPT-5.2 with enhanced reasoning. Does not support prompt caching.",
+          supportsTemperature: false
+        }
+      },
+      // GPT-5.1 Family
       {
         provider: "openai",
         modelId: "gpt-5.1",
         displayName: "GPT-5.1",
-        contextWindow: 128e3,
+        contextWindow: 1e6,
         maxOutputTokens: 32768,
         pricing: {
           input: 1.25,
@@ -5170,17 +5351,68 @@ var init_openai_models = __esm({
           fineTuning: true
         },
         metadata: {
-          family: "GPT-5",
+          family: "GPT-5.1",
           releaseDate: "2025-11-12",
-          notes: "Latest GPT-5 with improved instruction following. 2-3x faster than GPT-5.",
+          notes: "GPT-5 variant with improved instruction following. 2-3x faster than GPT-5.",
+          supportsTemperature: false
+        }
+      },
+      {
+        provider: "openai",
+        modelId: "gpt-5.1-codex",
+        displayName: "GPT-5.1 Codex",
+        contextWindow: 1e6,
+        maxOutputTokens: 32768,
+        pricing: {
+          input: 1.25,
+          output: 10,
+          cachedInput: 0.125
+        },
+        knowledgeCutoff: "2024-09-30",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "GPT-5.1",
+          notes: "GPT-5.1 variant optimized for code generation and analysis.",
+          supportsTemperature: false
+        }
+      },
+      {
+        provider: "openai",
+        modelId: "gpt-5.1-codex-max",
+        displayName: "GPT-5.1 Codex Max",
+        contextWindow: 1e6,
+        maxOutputTokens: 32768,
+        pricing: {
+          input: 1.25,
+          output: 10,
+          cachedInput: 0.125
+        },
+        knowledgeCutoff: "2024-09-30",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "GPT-5.1",
+          notes: "Extended thinking variant of GPT-5.1 Codex for complex code tasks.",
           supportsTemperature: false
         }
       },
+      // GPT-5 Family
       {
         provider: "openai",
         modelId: "gpt-5",
         displayName: "GPT-5",
-        contextWindow: 272e3,
+        contextWindow: 1e6,
         maxOutputTokens: 128e3,
         pricing: {
           input: 1.25,
@@ -5199,7 +5431,32 @@ var init_openai_models = __esm({
         metadata: {
           family: "GPT-5",
           releaseDate: "2025-08-07",
-          notes: "Best model for coding and agentic tasks. Adaptive reasoning with 90% caching discount.",
+          notes: "High-capability model for coding and agentic tasks. 90% caching discount.",
+          supportsTemperature: false
+        }
+      },
+      {
+        provider: "openai",
+        modelId: "gpt-5-codex",
+        displayName: "GPT-5 Codex",
+        contextWindow: 1e6,
+        maxOutputTokens: 128e3,
+        pricing: {
+          input: 1.25,
+          output: 10,
+          cachedInput: 0.125
+        },
+        knowledgeCutoff: "2024-09-30",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "GPT-5",
+          notes: "GPT-5 variant optimized for code generation and analysis.",
           supportsTemperature: false
         }
       },
@@ -5426,6 +5683,30 @@ var init_openai_models = __esm({
           supportsTemperature: false
         }
       },
+      {
+        provider: "openai",
+        modelId: "o1-pro",
+        displayName: "o1 Pro",
+        contextWindow: 2e5,
+        maxOutputTokens: 1e5,
+        pricing: {
+          input: 150,
+          output: 600
+        },
+        knowledgeCutoff: "2024-12-01",
+        features: {
+          streaming: true,
+          functionCalling: true,
+          vision: true,
+          reasoning: true,
+          structuredOutputs: true
+        },
+        metadata: {
+          family: "o-series",
+          notes: "Premium tier o1 with extended reasoning. Does not support prompt caching.",
+          supportsTemperature: false
+        }
+      },
       {
         provider: "openai",
         modelId: "o3",