diff --git a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_response.go b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_response.go
index af9ffef1..91bc0423 100644
--- a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_response.go
+++ b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_response.go
@@ -95,9 +95,9 @@ func ConvertAntigravityResponseToOpenAI(_ context.Context, _ string, originalReq
 		if totalTokenCountResult := usageResult.Get("totalTokenCount"); totalTokenCountResult.Exists() {
 			template, _ = sjson.Set(template, "usage.total_tokens", totalTokenCountResult.Int())
 		}
-		promptTokenCount := usageResult.Get("promptTokenCount").Int() - cachedTokenCount
+		promptTokenCount := usageResult.Get("promptTokenCount").Int()
 		thoughtsTokenCount := usageResult.Get("thoughtsTokenCount").Int()
-		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount+thoughtsTokenCount)
+		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount)
 		if thoughtsTokenCount > 0 {
 			template, _ = sjson.Set(template, "usage.completion_tokens_details.reasoning_tokens", thoughtsTokenCount)
 		}
diff --git a/internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go b/internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go
index 0415e014..b26d431f 100644
--- a/internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go
+++ b/internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go
@@ -100,7 +100,7 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
 		}
 		promptTokenCount := usageResult.Get("promptTokenCount").Int()
 		thoughtsTokenCount := usageResult.Get("thoughtsTokenCount").Int()
-		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount+thoughtsTokenCount)
+		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount)
 		if thoughtsTokenCount > 0 {
 			template, _ = sjson.Set(template, "usage.completion_tokens_details.reasoning_tokens", thoughtsTokenCount)
 		}
diff --git a/internal/translator/gemini/openai/chat-completions/gemini_openai_response.go b/internal/translator/gemini/openai/chat-completions/gemini_openai_response.go
index ee581c46..aeec5e9e 100644
--- a/internal/translator/gemini/openai/chat-completions/gemini_openai_response.go
+++ b/internal/translator/gemini/openai/chat-completions/gemini_openai_response.go
@@ -100,9 +100,9 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
 		if totalTokenCountResult := usageResult.Get("totalTokenCount"); totalTokenCountResult.Exists() {
 			baseTemplate, _ = sjson.Set(baseTemplate, "usage.total_tokens", totalTokenCountResult.Int())
 		}
-		promptTokenCount := usageResult.Get("promptTokenCount").Int() - cachedTokenCount
+		promptTokenCount := usageResult.Get("promptTokenCount").Int()
 		thoughtsTokenCount := usageResult.Get("thoughtsTokenCount").Int()
-		baseTemplate, _ = sjson.Set(baseTemplate, "usage.prompt_tokens", promptTokenCount+thoughtsTokenCount)
+		baseTemplate, _ = sjson.Set(baseTemplate, "usage.prompt_tokens", promptTokenCount)
 		if thoughtsTokenCount > 0 {
 			baseTemplate, _ = sjson.Set(baseTemplate, "usage.completion_tokens_details.reasoning_tokens", thoughtsTokenCount)
 		}
@@ -297,7 +297,7 @@ func ConvertGeminiResponseToOpenAINonStream(_ context.Context, _ string, origina
 		promptTokenCount := usageResult.Get("promptTokenCount").Int()
 		thoughtsTokenCount := usageResult.Get("thoughtsTokenCount").Int()
 		cachedTokenCount := usageResult.Get("cachedContentTokenCount").Int()
-		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount+thoughtsTokenCount)
+		template, _ = sjson.Set(template, "usage.prompt_tokens", promptTokenCount)
 		if thoughtsTokenCount > 0 {
 			template, _ = sjson.Set(template, "usage.completion_tokens_details.reasoning_tokens", thoughtsTokenCount)
 		}
diff --git a/internal/translator/gemini/openai/responses/gemini_openai-responses_response.go b/internal/translator/gemini/openai/responses/gemini_openai-responses_response.go
index 985897fa..73609be7 100644
--- a/internal/translator/gemini/openai/responses/gemini_openai-responses_response.go
+++ b/internal/translator/gemini/openai/responses/gemini_openai-responses_response.go
@@ -531,8 +531,8 @@ func ConvertGeminiResponseToOpenAIResponses(_ context.Context, modelName string,
 
 		// usage mapping
 		if um := root.Get("usageMetadata"); um.Exists() {
-			// input tokens = prompt + thoughts
-			input := um.Get("promptTokenCount").Int() + um.Get("thoughtsTokenCount").Int()
+			// input tokens = prompt only (thoughts go to output)
+			input := um.Get("promptTokenCount").Int()
 			completed, _ = sjson.Set(completed, "response.usage.input_tokens", input)
 			// cached token details: align with OpenAI "cached_tokens" semantics.
 			completed, _ = sjson.Set(completed, "response.usage.input_tokens_details.cached_tokens", um.Get("cachedContentTokenCount").Int())
@@ -737,8 +737,8 @@ func ConvertGeminiResponseToOpenAIResponsesNonStream(_ context.Context, _ string
 
 	// usage mapping
 	if um := root.Get("usageMetadata"); um.Exists() {
-		// input tokens = prompt + thoughts
-		input := um.Get("promptTokenCount").Int() + um.Get("thoughtsTokenCount").Int()
+		// input tokens = prompt only (thoughts go to output)
+		input := um.Get("promptTokenCount").Int()
 		resp, _ = sjson.Set(resp, "usage.input_tokens", input)
 		// cached token details: align with OpenAI "cached_tokens" semantics.
 		resp, _ = sjson.Set(resp, "usage.input_tokens_details.cached_tokens", um.Get("cachedContentTokenCount").Int())