diff --git a/nanobot/providers/anthropic_provider.py b/nanobot/providers/anthropic_provider.py
index 3c789e730..fabcd5656 100644
--- a/nanobot/providers/anthropic_provider.py
+++ b/nanobot/providers/anthropic_provider.py
@@ -379,6 +379,10 @@ class AnthropicProvider(LLMProvider):
                 val = getattr(response.usage, attr, 0)
                 if val:
                     usage[attr] = val
+            # Normalize to cached_tokens for downstream consistency.
+            cache_read = usage.get("cache_read_input_tokens", 0)
+            if cache_read:
+                usage["cached_tokens"] = cache_read
 
         return LLMResponse(
             content="".join(content_parts) or None,
diff --git a/tests/providers/test_cached_tokens.py b/tests/providers/test_cached_tokens.py
index 8bc2555d1..fce22cf65 100644
--- a/tests/providers/test_cached_tokens.py
+++ b/tests/providers/test_cached_tokens.py
@@ -189,3 +189,43 @@ def test_extract_usage_priority_nested_over_top_level_dict():
     }
     result = p._parse(response)
     assert result.usage["cached_tokens"] == 100
+
+
+def test_anthropic_maps_cache_fields_to_cached_tokens():
+    """Anthropic's cache_read_input_tokens should map to cached_tokens."""
+    from nanobot.providers.anthropic_provider import AnthropicProvider
+
+    usage_obj = FakeUsage(
+        input_tokens=800,
+        output_tokens=200,
+        cache_creation_input_tokens=0,
+        cache_read_input_tokens=1200,
+    )
+    content_block = FakeUsage(type="text", text="hello")
+    response = FakeUsage(
+        id="msg_1",
+        type="message",
+        stop_reason="end_turn",
+        content=[content_block],
+        usage=usage_obj,
+    )
+    result = AnthropicProvider._parse_response(response)
+    assert result.usage["cached_tokens"] == 1200
+    assert result.usage["prompt_tokens"] == 800
+
+
+def test_anthropic_no_cache_fields():
+    """Anthropic response without cache fields should not have cached_tokens."""
+    from nanobot.providers.anthropic_provider import AnthropicProvider
+
+    usage_obj = FakeUsage(input_tokens=800, output_tokens=200)
+    content_block = FakeUsage(type="text", text="hello")
+    response = FakeUsage(
+        id="msg_1",
+        type="message",
+        stop_reason="end_turn",
+        content=[content_block],
+        usage=usage_obj,
+    )
+    result = AnthropicProvider._parse_response(response)
+    assert "cached_tokens" not in result.usage