diff --git a/.changeset/exclude-chatgpt-from-prompt-cache-breakpoint.md b/.changeset/exclude-chatgpt-from-prompt-cache-breakpoint.md
new file mode 100644
index 00000000000..6362bdd7ff1
--- /dev/null
+++ b/.changeset/exclude-chatgpt-from-prompt-cache-breakpoint.md
@@ -0,0 +1,5 @@
+---
+"@kilocode/cli": patch
+---
+
+Exclude ChatGPT subscriptions from explicit prompt cache breakpoints.
diff --git a/packages/opencode/src/provider/transform.ts b/packages/opencode/src/provider/transform.ts
index ff534bf107f..310c9d0c83c 100644
--- a/packages/opencode/src/provider/transform.ts
+++ b/packages/opencode/src/provider/transform.ts
@@ -328,15 +328,20 @@ function normalizeMessages(
   return msgs
 }
 
-// kilocode_change start - explicit prompt cache breakpoints for GPT-5.6+
-function supportsPromptCacheBreakpoint(modelId: string): boolean {
-  const match = modelId.match(/gpt-(\d+)\.(\d+)/)
+// kilocode_change start - explicit prompt cache breakpoints for GPT-5.6+ (excluding ChatGPT subscriptions)
+function isLikelyChatGPTSubscription(model: Provider.Model): boolean {
+  return model.providerID === "openai" && model.cost?.input === 0 && model.cost?.output === 0
+}
+
+function supportsPromptCacheBreakpoint(model: Provider.Model): boolean {
+  if (isLikelyChatGPTSubscription(model)) return false
+  const match = model.api.id.match(/gpt-(\d+)\.(\d+)/)
   if (match) {
     const major = Number(match[1])
     const minor = Number(match[2])
     if (major > 5 || (major === 5 && minor >= 6)) return true
   }
-  const majorMatch = modelId.match(/gpt-(\d+)/)
+  const majorMatch = model.api.id.match(/gpt-(\d+)/)
   if (majorMatch && Number(majorMatch[1]) >= 6) return true
   return false
 }
@@ -366,7 +371,7 @@ function applyCaching(msgs: ModelMessage[], model: Provider.Model): ModelMessage
       cacheControl: { type: "ephemeral" },
     },
     // kilocode_change start
-    ...(supportsPromptCacheBreakpoint(model.api.id)
+    ...(supportsPromptCacheBreakpoint(model)
       ? {
           openai: {
             promptCacheBreakpoint: { mode: "explicit" },
@@ -494,7 +499,7 @@ export function message(msgs: ModelMessage[], model: Provider.Model, options: Re
       ((model.api.npm === "@ai-sdk/openai" ||
         model.api.npm === "@ai-sdk/azure" ||
         model.api.npm === "@kilocode/kilo-gateway") &&
-        supportsPromptCacheBreakpoint(model.api.id))) &&
+        supportsPromptCacheBreakpoint(model))) &&
     model.api.npm !== "@ai-sdk/gateway"
   ) {
     msgs = applyCaching(msgs, model)
diff --git a/packages/opencode/test/provider/transform.test.ts b/packages/opencode/test/provider/transform.test.ts
index 74ef174ade8..e834ec326eb 100644
--- a/packages/opencode/test/provider/transform.test.ts
+++ b/packages/opencode/test/provider/transform.test.ts
@@ -3162,6 +3162,38 @@ describe("ProviderTransform.message - cache control on gateway", () => {
       },
     })
   })
+
+  test("openai gpt-5.6 with ChatGPT subscription (zero cost heuristic) does not apply promptCacheBreakpoint", () => {
+    const model = createModel({
+      providerID: "openai",
+      api: {
+        id: "gpt-5.6",
+        url: "https://api.openai.com/v1",
+        npm: "@ai-sdk/openai",
+      },
+      id: "gpt-5.6",
+      cost: {
+        input: 0,
+        output: 0,
+        cache: { read: 0, write: 0 },
+      },
+    })
+    const msgs = [
+      {
+        role: "system",
+        content: "You are a helpful assistant",
+      },
+      {
+        role: "user",
+        content: "Hello",
+      },
+    ] as any[]
+
+    const result = ProviderTransform.message(msgs, model, {}) as any[]
+
+    expect(result[0].providerOptions?.openai?.promptCacheBreakpoint).toBeUndefined()
+    expect(result[1].providerOptions?.openai?.promptCacheBreakpoint).toBeUndefined()
+  })
   // kilocode_change end
 })