diff --git a/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.test.tsx b/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.test.tsx
index f31e1839739..4ba6d8d50b4 100644
--- a/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.test.tsx
+++ b/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.test.tsx
@@ -51,4 +51,23 @@ describe("ResponseMetrics prompt cache chips", () => {
 
     expect(screen.queryByText(/Response Cache/)).not.toBeInTheDocument();
   });
+
+  it("does not render the Cost chip when a persisted cost is null", () => {
+    render(<ResponseMetrics usage={{ promptTokens: 1, cost: null as unknown as number }} />);
+
+    expect(screen.queryByText(/Cost:/)).not.toBeInTheDocument();
+    expect(screen.getByText("In: 1")).toBeInTheDocument();
+  });
+
+  it("does not render the Cost chip for NaN", () => {
+    render(<ResponseMetrics usage={{ ...baseUsage, cost: Number.NaN }} />);
+
+    expect(screen.queryByText(/Cost:/)).not.toBeInTheDocument();
+  });
+
+  it("renders the Cost chip for a finite cost", () => {
+    render(<ResponseMetrics usage={{ ...baseUsage, cost: 0.000063 }} />);
+
+    expect(screen.getByText("Cost: $0.000063")).toBeInTheDocument();
+  });
 });
diff --git a/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.tsx b/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.tsx
index ec62d0618d7..bb7debc7aee 100644
--- a/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.tsx
+++ b/ui/litellm-dashboard/src/components/chat_ui/ResponseMetrics.tsx
@@ -159,7 +159,7 @@ const ResponseMetrics: React.FC<ResponseMetricsProps> = ({ timeToFirstToken, tot
         />
       )}
 
-      {usage?.cost !== undefined && (
+      {typeof usage?.cost === "number" && Number.isFinite(usage.cost) && (
         <MetricItem
           label="Cost"
           tooltip="Cost"
diff --git a/ui/litellm-dashboard/src/components/llm_calls/chat_completion.test.tsx b/ui/litellm-dashboard/src/components/llm_calls/chat_completion.test.tsx
index bc4b4e3a351..129d51d50a2 100644
--- a/ui/litellm-dashboard/src/components/llm_calls/chat_completion.test.tsx
+++ b/ui/litellm-dashboard/src/components/llm_calls/chat_completion.test.tsx
@@ -466,6 +466,12 @@ describe("chat_completion prompt cache usage", () => {
     expect(usageData).not.toHaveProperty("cacheReadTokens");
     expect(usageData).not.toHaveProperty("cacheCreationTokens");
   });
+
+  it("omits cost when the provider reports a non-numeric value", async () => {
+    const usageData = await captureUsage({ cost: "not-a-number" });
+
+    expect(usageData).toEqual(expect.not.objectContaining({ cost: expect.anything() }));
+  });
 });
 
 describe("chat_completion response cache", () => {
diff --git a/ui/litellm-dashboard/src/components/llm_calls/chat_completion.tsx b/ui/litellm-dashboard/src/components/llm_calls/chat_completion.tsx
index cd0852bc06e..b813a35f687 100644
--- a/ui/litellm-dashboard/src/components/llm_calls/chat_completion.tsx
+++ b/ui/litellm-dashboard/src/components/llm_calls/chat_completion.tsx
@@ -245,7 +245,10 @@ export async function makeOpenAIChatCompletionRequest(
 
         // Extract cost from usage object if available
         if (chunkWithUsage.usage.cost !== undefined && chunkWithUsage.usage.cost !== null) {
-          usageData.cost = parseFloat(chunkWithUsage.usage.cost);
+          const parsedCost = parseFloat(chunkWithUsage.usage.cost);
+          if (Number.isFinite(parsedCost)) {
+            usageData.cost = parsedCost;
+          }
         }
 
         onUsageData(usageData);
diff --git a/ui/litellm-dashboard/src/components/llm_calls/responses_api.test.tsx b/ui/litellm-dashboard/src/components/llm_calls/responses_api.test.tsx
index 0b94c093acd..7b13f0ac2e3 100644
--- a/ui/litellm-dashboard/src/components/llm_calls/responses_api.test.tsx
+++ b/ui/litellm-dashboard/src/components/llm_calls/responses_api.test.tsx
@@ -231,6 +231,35 @@ describe("responses_api", () => {
     expect(onUsageData).toHaveBeenCalledWith(expect.not.objectContaining({ cost: expect.anything() }), "");
   });
 
+  it("should omit cost when the proxy reports a non-numeric cost", async () => {
+    async function* streamWithNonNumericCost() {
+      yield {
+        type: "response.completed",
+        response: {
+          id: "resp_non_numeric_cost",
+          usage: { output_tokens: 12, input_tokens: 12, total_tokens: 24, cost: "not-a-number" },
+        },
+      };
+    }
+    mockResponsesCreate.mockResolvedValueOnce(streamWithNonNumericCost());
+
+    const onUsageData = vi.fn();
+
+    await makeOpenAIResponsesRequest(
+      messages,
+      mockUpdateTextUI,
+      "gpt-4",
+      "test-token",
+      undefined,
+      undefined,
+      undefined,
+      undefined,
+      onUsageData,
+    );
+
+    expect(onUsageData).toHaveBeenCalledWith(expect.not.objectContaining({ cost: expect.anything() }), "");
+  });
+
   it("should replay MCP output items as events for a non-streaming response", async () => {
     mockResponsesCreate.mockReturnValueOnce(
       nonStreamingResponse({
diff --git a/ui/litellm-dashboard/src/components/llm_calls/responses_api.tsx b/ui/litellm-dashboard/src/components/llm_calls/responses_api.tsx
index 7ab76488504..63085a81b94 100644
--- a/ui/litellm-dashboard/src/components/llm_calls/responses_api.tsx
+++ b/ui/litellm-dashboard/src/components/llm_calls/responses_api.tsx
@@ -312,7 +312,10 @@ export async function makeOpenAIResponsesRequest(
             }
 
             if (usage.cost !== undefined && usage.cost !== null) {
-              usageData.cost = Number(usage.cost);
+              const parsedCost = Number(usage.cost);
+              if (Number.isFinite(parsedCost)) {
+                usageData.cost = parsedCost;
+              }
             }
 
             onUsageData(usageData, mcpToolUsed);