Skip to content

7.4 billion tokens per month. 34 free LLM providers. 635 free model endpoints. All behind one /v1 endpoint, plus any custom OpenAI-compatible endpoint. Smart routing, automatic failover, encrypted keys. Personal experimentation only.

TypeScript

fix(apinex): enforce the shared 5 RPM free-tier cap across all models (#1584)

Tashfeen0c456ea4df99
diff --git a/server/src/__tests__/services/provider-minute-cap.test.ts b/server/src/__tests__/services/provider-minute-cap.test.ts
index 0bb1b4fb..053d9d19 100644
--- a/server/src/__tests__/services/provider-minute-cap.test.ts
+++ b/server/src/__tests__/services/provider-minute-cap.test.ts
@@ -31,6 +31,7 @@ describe('provider-wide per-minute request cap', () => {
afterEach(() => {
delete process.env.PROVIDER_MINUTE_REQUEST_CAP_NVIDIA;
delete process.env.PROVIDER_MINUTE_REQUEST_CAP_GROQ;
+ delete process.env.PROVIDER_MINUTE_REQUEST_CAP_APINEX;
});
it('ships a default cap for nvidia and none for a provider without one', () => {
@@ -38,6 +39,17 @@ describe('provider-wide per-minute request cap', () => {
expect(getProviderMinuteRequestCap('groq')).toBeNull();
});
+ it('applies the APINex 5 RPM cap across different free models, per key', () => {
+ const now = Date.now();
+ expect(getProviderMinuteRequestCap('apinex')).toBe(5);
+ seedRequests('apinex', 'free/deepseek-v4.1-flash', keyId, 3, now - 20_000);
+ seedRequests('apinex', 'free/gpt-6-luna', keyId, 2, now - 10_000);
+ expect(providerMinuteRequestCount('apinex', keyId, now)).toBe(5);
+ expect(canUseProviderMinute('apinex', keyId, now)).toBe(false);
+ expect(canUseProviderMinute('apinex', keyId + 1, now)).toBe(true);
+ expect(canUseProviderMinute('apinex', keyId, now + 61_000)).toBe(true);
+ });
+
it('honours an env override and treats 0 as disabled', () => {
process.env.PROVIDER_MINUTE_REQUEST_CAP_NVIDIA = '5';
expect(getProviderMinuteRequestCap('nvidia')).toBe(5);
diff --git a/server/src/services/ratelimit.ts b/server/src/services/ratelimit.ts
index 06dc5156..51f40c3b 100644
--- a/server/src/services/ratelimit.ts
+++ b/server/src/services/ratelimit.ts
@@ -587,6 +587,9 @@ const DEFAULT_PROVIDER_DAILY_TOKEN_CAPS: Record<string, number> = {
// entirely — so it keeps dispatching and eats real 429s.
const DEFAULT_PROVIDER_MINUTE_REQUEST_CAPS: Record<string, number> = {
nvidia: 40,
+ // APINex Free: 5 requests/minute per key shared across every model
+ // (apinex.bond/api/public/rate-limits `rpmFree`), not 5 per model.
+ apinex: 5,
};
export function getProviderDailyRequestCap(platform: string): number | null {