Add Anthropic prompt caching (direct + via OpenRouter)

Caches the system prompt/tools and growing conversation history via
cache_control breakpoints, cutting cost and latency on repeated turns.
Covers both the regular chat path and the tool-calling loop
(chatWithToolMessages), which has its own request-building code and was
initially missed. Cost calculation now accounts for cache write/read
pricing instead of treating all input tokens as full price. Verified
live: cache reads grow turn-over-turn in oAI.log.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

oAI/Providers/OpenRouterProvider.swift

@@ -198,6 +198,11 @@ class OpenRouterProvider: AIProvider {
         }
         if let maxTokens = maxTokens { body["max_tokens"] = maxTokens }
         if let temperature = temperature { body["temperature"] = temperature }
+        // Anthropic models require an explicit cache_control opt-in on OpenRouter;
+        // other providers cache automatically.
+        if model.hasPrefix("anthropic/") {
+            body["cache_control"] = ["type": "ephemeral"]
+        }
 
         var urlRequest = URLRequest(url: url)
         urlRequest.httpMethod = "POST"
@@ -388,6 +393,12 @@ class OpenRouterProvider: AIProvider {
             ReasoningAPIConfig(effort: $0.effort, exclude: $0.exclude ? true : nil)
         }
 
+        // Anthropic models require an explicit cache_control opt-in on OpenRouter;
+        // other providers (OpenAI, DeepSeek, Gemini, Grok, etc.) cache automatically.
+        let cacheControl: OpenRouterChatRequest.CacheControl? = effectiveModel.hasPrefix("anthropic/")
+            ? .init(type: "ephemeral")
+            : nil
+
         return OpenRouterChatRequest(
             model: effectiveModel,
             messages: apiMessages,
@@ -398,7 +409,8 @@ class OpenRouterProvider: AIProvider {
             tools: request.tools,
             toolChoice: request.tools != nil ? "auto" : nil,
             modalities: request.imageGeneration ? ["text", "image"] : nil,
-            reasoning: reasoningConfig
+            reasoning: reasoningConfig,
+            cacheControl: cacheControl
         )
     }
     
@@ -416,6 +428,11 @@ class OpenRouterProvider: AIProvider {
         let allImages = topLevelImages + blockImages
         let images: [Data]? = allImages.isEmpty ? nil : allImages
 
+        if let details = apiResponse.usage?.promptTokensDetails,
+           details.cachedTokens != nil || details.cacheWriteTokens != nil {
+            Log.api.info("OpenRouter cache usage: model=\(apiResponse.model), created=\(details.cacheWriteTokens ?? 0), read=\(details.cachedTokens ?? 0)")
+        }
+
         return ChatResponse(
             id: apiResponse.id,
             model: apiResponse.model,
@@ -426,7 +443,9 @@ class OpenRouterProvider: AIProvider {
                 ChatResponse.Usage(
                     promptTokens: usage.promptTokens,
                     completionTokens: usage.completionTokens,
-                    totalTokens: usage.totalTokens
+                    totalTokens: usage.totalTokens,
+                    cacheCreationInputTokens: usage.promptTokensDetails?.cacheWriteTokens,
+                    cacheReadInputTokens: usage.promptTokensDetails?.cachedTokens
                 )
             },
             created: Date(timeIntervalSince1970: TimeInterval(apiResponse.created)),
@@ -446,6 +465,11 @@ class OpenRouterProvider: AIProvider {
         let allImages = topLevelImages + blockImages
         let images: [Data]? = allImages.isEmpty ? nil : allImages
 
+        if let details = apiChunk.usage?.promptTokensDetails,
+           details.cachedTokens != nil || details.cacheWriteTokens != nil {
+            Log.api.info("OpenRouter stream cache usage: model=\(apiChunk.model), created=\(details.cacheWriteTokens ?? 0), read=\(details.cachedTokens ?? 0)")
+        }
+
         return StreamChunk(
             id: apiChunk.id,
             model: apiChunk.model,
@@ -460,7 +484,9 @@ class OpenRouterProvider: AIProvider {
                 ChatResponse.Usage(
                     promptTokens: usage.promptTokens,
                     completionTokens: usage.completionTokens,
-                    totalTokens: usage.totalTokens
+                    totalTokens: usage.totalTokens,
+                    cacheCreationInputTokens: usage.promptTokensDetails?.cacheWriteTokens,
+                    cacheReadInputTokens: usage.promptTokensDetails?.cachedTokens
                 )
             }
         )