fix(pricing): price the NVIDIA free catalog models at zero

changelog.d/fixes/18464-nvidia-missing-pricing-rows.md

@@ -0,0 +1 @@
+- **fix(pricing):** add zero-price rows for the seven NVIDIA catalog models that had none (`meta/muse-glimmer-30b`, `poolside/laguna-xs-2.1`, `google/gemma-4-31b-it`, `google/diffusiongemma-26b-a4b-it`, `nvidia/nemotron-3-ultra-550b-a55b`, `nvidia/nemotron-3.5-lightning-30b-a3b`, `nvidia/nemotron-3-nano-omni-30b-a3b-reasoning`). They run on NVIDIA's free hosted endpoint, which has no per-token price listed, but without a row their usage was unpriced and an API key with a USD limit failed closed after one request ([#18464](https://github.com/diegosouzapw/OmniRoute/issues/18464))

src/shared/constants/pricing/inference-hosts.ts

@@ -380,6 +380,40 @@ export const DEFAULT_PRICING_INFERENCE = {
     "nvidia/gpt-oss-120b": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
     "openai/gpt-oss-120b": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
     "gpt-oss-120b": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
+    // #18464: NVIDIA hosted free endpoints. build.nvidia.com rate limits them and lists no
+    // per-token price, so they are priced at zero. Without a row, usage is unpriced and
+    // USD-limited API keys fail closed (#12341) after one request.
+    "meta/muse-glimmer-30b": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
+    "poolside/laguna-xs-2.1": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
+    "google/gemma-4-31b-it": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },
+    "google/diffusiongemma-26b-a4b-it": {
+      input: 0,
+      output: 0,
+      cached: 0,
+      reasoning: 0,
+      cache_creation: 0,
+    },
+    "nvidia/nemotron-3-ultra-550b-a55b": {
+      input: 0,
+      output: 0,
+      cached: 0,
+      reasoning: 0,
+      cache_creation: 0,
+    },
+    "nvidia/nemotron-3.5-lightning-30b-a3b": {
+      input: 0,
+      output: 0,
+      cached: 0,
+      reasoning: 0,
+      cache_creation: 0,
+    },
+    "nvidia/nemotron-3-nano-omni-30b-a3b-reasoning": {
+      input: 0,
+      output: 0,
+      cached: 0,
+      reasoning: 0,
+      cache_creation: 0,
+    },
   },
   nebius: {
     "openai/gpt-oss-120b": { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 },

tests/unit/nvidia-registry-pricing-18464.test.ts

@@ -0,0 +1,47 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { nvidiaProvider } from "../../open-sse/config/providers/registry/nvidia/index.ts";
+import { calculateCostDetailed } from "../../src/lib/usage/costCalculator.ts";
+import { getDefaultPricing } from "../../src/shared/constants/pricing.ts";
+
+// #18464: every NVIDIA registry model runs on NVIDIA's free hosted endpoint, which is
+// rate limited and has no per-token price listed. A registry model without a pricing row is
+// reported unpriced, and an API key with a USD limit then fails closed (#12341)
+// after a single request to it.
+//
+// Ids that the fix for #18418 handles instead of this one. It removes the
+// first three from the registry, because NVIDIA retired them (410 end of life), and
+// adds the zero row for the fourth. Once that fix is merged, this list can be emptied.
+const HANDLED_BY_18418 = new Set<string>([
+  "moonshotai/kimi-k3",
+  "deepseek-ai/deepseek-v4-pro-0813",
+  "deepseek-ai/deepseek-v4-flash-0731",
+  "nvidia/nemotron-3.5-super-vl-120b-a12b",
+]);
+
+const ZERO_ROW = { input: 0, output: 0, cached: 0, reasoning: 0, cache_creation: 0 };
+
+const checkedIds = nvidiaProvider.models
+  .map((model) => model.id)
+  .filter((id) => !HANDLED_BY_18418.has(id));
+
+test("every NVIDIA registry model has a zero-price pricing row (#18464, #12341)", async () => {
+  const pricing = getDefaultPricing() as Record<string, Record<string, Record<string, number>>>;
+  assert.ok(checkedIds.length > 0, "the nvidia registry walk must check at least one model");
+  const unpriced: string[] = [];
+  for (const id of checkedIds) {
+    const result = await calculateCostDetailed("nvidia", id, { input: 1000, output: 1000 });
+    if (!result.priced) {
+      unpriced.push(id);
+      continue;
+    }
+    assert.equal(result.costUsd, 0, `nvidia/${id} is a free endpoint and must cost 0`);
+    assert.deepEqual(pricing.nvidia?.[id], ZERO_ROW, `nvidia/${id} needs an all zero row`);
+  }
+  assert.deepEqual(
+    unpriced,
+    [],
+    `NVIDIA registry models without a pricing row: ${unpriced.join(", ")}`
+  );
+});