diff --git a/cli/mcp_metering.py b/cli/mcp_metering.py
index 2851d49..2f3f466 100644
--- a/cli/mcp_metering.py
+++ b/cli/mcp_metering.py
@@ -22,6 +22,10 @@ def _account_id() -> str:
     return os.environ.get("NUNCHI_ACCOUNT_ID", "").strip()
 
 
+def _user_id() -> str:
+    return os.environ.get("NUNCHI_USER_ID", "").strip()
+
+
 def _subscription_id() -> str:
     return os.environ.get("NUNCHI_SUBSCRIPTION_ID", "").strip()
 
@@ -72,6 +76,8 @@ def upload_rows(rows: list[dict[str, Any]]) -> dict[str, Any]:
     body = {
         "account_id": account_id,
         "accountId": account_id,
+        "user_id": _user_id() or None,
+        "userId": _user_id() or None,
         "subscription_id": _subscription_id() or None,
         "subscriptionId": _subscription_id() or None,
         "plan_id": _plan_id(),
@@ -118,6 +124,11 @@ def report_inference_cost(
     cached_tokens: Optional[int] = None,
     cache_savings_usd: Optional[Any] = None,
     model: Optional[str] = None,
+    provider: Optional[str] = None,
+    requested_model: Optional[str] = None,
+    route: Optional[str] = None,
+    decision_call_id: Optional[str] = None,
+    price_source: Optional[str] = None,
 ) -> dict[str, Any]:
     if not metering_enabled():
         return {"ok": False, "skipped": "metering_not_configured"}
@@ -128,12 +139,28 @@ def report_inference_cost(
         "input_tokens": int(input_tokens or 0),
         "output_tokens": int(output_tokens or 0),
         "ts": ts,
+        "account_id": _account_id(),
+        "subscription_id": _subscription_id() or None,
+        "plan_id": _plan_id(),
     }
+    if _user_id():
+        row["user_id"] = _user_id()
     if cached_tokens is not None:
         row["cached_tokens"] = int(cached_tokens)
     if cache_savings_usd is not None:
         row["cache_savings_usd"] = cache_savings_usd
     if model:
         row["model"] = model
+        row["resolved_model"] = model
+    if provider:
+        row["provider"] = provider
+    if requested_model:
+        row["requested_model"] = requested_model
+    if route:
+        row["route"] = route
+    if decision_call_id:
+        row["decision_call_id"] = decision_call_id
+    if price_source:
+        row["pricing_snapshot_source"] = price_source
     row_id = _row_id("cost", row)
     return upload_rows([{"row_id": row_id, "metric_type": "cost", "row": row}])
diff --git a/modules/cost_metering.py b/modules/cost_metering.py
index a1c5879..d4023cf 100644
--- a/modules/cost_metering.py
+++ b/modules/cost_metering.py
@@ -297,6 +297,11 @@ def record_llm_call(
                 cached_tokens=cached_tokens,
                 cache_savings_usd=str(cache_savings) if cache_savings is not None else None,
                 model=resolved_model,
+                provider=provider,
+                requested_model=requested_model,
+                route=route,
+                decision_call_id=decision_call_id,
+                price_source=row["pricing_snapshot_source"],
             )
         except Exception:
             pass
diff --git a/tests/test_mcp_metering.py b/tests/test_mcp_metering.py
index a4691d8..96732fd 100644
--- a/tests/test_mcp_metering.py
+++ b/tests/test_mcp_metering.py
@@ -1,4 +1,4 @@
-from cli.mcp_metering import _tool_bucket, report_tool_call, upload_rows
+from cli.mcp_metering import _tool_bucket, report_inference_cost, report_tool_call, upload_rows
 
 
 def test_tool_bucket():
@@ -42,3 +42,58 @@ def fake_urlopen(req, timeout=10):
     body = __import__("json").loads(captured["body"])
     assert body["accountId"] == "acct_1"
     assert body["rows"][0]["metric_type"] == "mcp_tool"
+
+
+def test_report_inference_cost_includes_subscription_identity(monkeypatch):
+    captured = {}
+
+    class FakeResponse:
+        def read(self):
+            return b'{"ok": true}'
+
+        def __enter__(self):
+            return self
+
+        def __exit__(self, exc_type, exc, tb):
+            return False
+
+    def fake_urlopen(req, timeout=10):
+        captured["body"] = req.data.decode("utf-8")
+        return FakeResponse()
+
+    monkeypatch.setenv("NUNCHI_METERING_URL", "https://pair.example/api/metering/usage")
+    monkeypatch.setenv("NUNCHI_METERING_TOKEN", "token")
+    monkeypatch.setenv("NUNCHI_USER_ID", "user_1")
+    monkeypatch.setenv("NUNCHI_ACCOUNT_ID", "acct_1")
+    monkeypatch.setenv("NUNCHI_SUBSCRIPTION_ID", "sub_1")
+    monkeypatch.setenv("NUNCHI_PLAN_ID", "hosted-mcp-inference-starter")
+    monkeypatch.setattr("cli.mcp_metering.urllib.request.urlopen", fake_urlopen)
+
+    result = report_inference_cost(
+        inference_usd="0.0042",
+        input_tokens=12,
+        output_tokens=2,
+        cached_tokens=3,
+        cache_savings_usd="0.0001",
+        model="openai/gpt-4.1-mini",
+        provider="openrouter",
+        requested_model="openrouter/auto",
+        route="openrouter/auto",
+        decision_call_id="decision-1",
+        price_source="openrouter:usage.cost",
+    )
+
+    assert result["ok"] is True
+    body = __import__("json").loads(captured["body"])
+    row = body["rows"][0]["row"]
+    assert body["accountId"] == "acct_1"
+    assert body["subscriptionId"] == "sub_1"
+    assert row["user_id"] == "user_1"
+    assert row["account_id"] == "acct_1"
+    assert row["subscription_id"] == "sub_1"
+    assert row["plan_id"] == "hosted-mcp-inference-starter"
+    assert row["inference_usd"] == "0.0042"
+    assert row["provider"] == "openrouter"
+    assert row["requested_model"] == "openrouter/auto"
+    assert row["resolved_model"] == "openai/gpt-4.1-mini"
+    assert row["decision_call_id"] == "decision-1"