diff --git a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request.go b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request.go
index f03020a5..975e6479 100644
--- a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request.go
+++ b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request.go
@@ -59,7 +59,7 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 	}
 	out = applyOpenAIThinkingCompatibilityToAntigravity(out, rawJSON)
 
-	// Temperature/top_p/top_k/max_tokens
+	// Temperature/top_p/top_k/max_tokens/max_completion_tokens
 	if tr := gjson.GetBytes(rawJSON, "temperature"); tr.Exists() && tr.Type == gjson.Number {
 		out, _ = sjson.SetBytes(out, "request.generationConfig.temperature", tr.Num)
 	}
@@ -71,6 +71,8 @@ func ConvertOpenAIRequestToAntigravity(modelName string, inputRawJSON []byte, _
 	}
 	if maxTok := gjson.GetBytes(rawJSON, "max_tokens"); maxTok.Exists() && maxTok.Type == gjson.Number {
 		out, _ = sjson.SetBytes(out, "request.generationConfig.maxOutputTokens", maxTok.Num)
+	} else if mct := gjson.GetBytes(rawJSON, "max_completion_tokens"); mct.Exists() && mct.Type == gjson.Number {
+		out, _ = sjson.SetBytes(out, "request.generationConfig.maxOutputTokens", mct.Num)
 	}
 
 	// Map OpenAI response_format to Antigravity structured output settings.
diff --git a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request_test.go b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request_test.go
index ac884848..5d9a649d 100644
--- a/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request_test.go
+++ b/internal/translator/antigravity/openai/chat-completions/antigravity_openai_request_test.go
@@ -455,3 +455,40 @@ func TestConvertOpenAIRequestToAntigravityTranslatesVideoURL(t *testing.T) {
 		t.Fatalf("inlineData.data = %q, want AAAAIGZ0eXBtcDQy. Output: %s", got, out)
 	}
 }
+
+func TestConvertOpenAIRequestToAntigravity_MaxCompletionTokens(t *testing.T) {
+	tests := []struct {
+		name     string
+		body     string
+		expected float64
+	}{
+		{
+			name:     "only max_tokens",
+			body:     `{"model":"gemini-2.5-flash","messages":[{"role":"user","content":"hi"}],"max_tokens":100}`,
+			expected: 100,
+		},
+		{
+			name:     "only max_completion_tokens",
+			body:     `{"model":"gemini-2.5-flash","messages":[{"role":"user","content":"hi"}],"max_completion_tokens":200}`,
+			expected: 200,
+		},
+		{
+			name:     "max_tokens preferred over max_completion_tokens",
+			body:     `{"model":"gemini-2.5-flash","messages":[{"role":"user","content":"hi"}],"max_tokens":100,"max_completion_tokens":200}`,
+			expected: 100,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			out := ConvertOpenAIRequestToAntigravity("gemini-2.5-flash", []byte(tt.body), false)
+			got := gjson.GetBytes(out, "request.generationConfig.maxOutputTokens")
+			if !got.Exists() {
+				t.Fatalf("request.generationConfig.maxOutputTokens missing. Output: %s", out)
+			}
+			if got.Float() != tt.expected {
+				t.Fatalf("maxOutputTokens = %v, want %v. Output: %s", got.Float(), tt.expected, out)
+			}
+		})
+	}
+}