Billing & Usage

HX-SDP meters all API operations in Compute Units (CUs). Each tenant has a monthly CU quota determined by their billing tier.

---

CU cost table

Operation	Endpoint	CU cost	Description
put	POST /v1/put	1.0	Ingest + QTT-SVD compression
put_cores	POST /v1/put_cores	1.0	Direct TT-core upload
put_cores_batch	POST /v1/put_cores_batch	2.0	Batch TT-core upload
get	GET /v1/get/{ns}/{key}	0.1	Metadata retrieval (no decompress)
serve	POST /v1/serve (dense)	0.5	Dense array reconstruction
serve_gpu	POST /v1/serve (gpu)	2.0	Load TT-cores to GPU VRAM
search	POST /v1/search	0.5	Metadata-filtered search
delete	DELETE /v1/delete/{ns}/{key}	0.1	Soft-delete entry
list	GET /v1/list/{ns}	0.1	List keys in namespace
query_similarity	POST /v1/query/similarity	1.0	Pairwise core inner product
query_topk	POST /v1/query/topk	1.0	Top-K similarity search
query_vector	POST /v1/query/vector	1.0	External vector top-K search

The X-HX-CUs response header reports the cost charged for each request.

---

Billing tiers

Managed (SaaS)

Tier	Price	Monthly CUs	Rate limit	Overage
Starter	Free	1,000	10 req/min	—
Builder	$49/mo	50,000	100 req/60s	$0.05/CU
Pro	$299/mo	500,000	1,000 req/60s	$0.03/CU
Enterprise	Custom	Custom	Custom	Custom

Self-hosted

Self-hosted deployments have no CU limits by default. Enable metering via HX_GATE_BILLING_ENABLED=true if you want to track usage across internal tenants.

---

Checking usage

Via API

# All tenants (admin)
curl https://gate.holonomx.com/gate/admin/usage \
  -H "Authorization: Bearer $SERVICE_KEY"

# Single tenant (admin)
curl https://gate.holonomx.com/gate/admin/usage/acme-corp \
  -H "Authorization: Bearer $SERVICE_KEY"

Via Console

The Console dashboard shows a real-time usage bar for the authenticated tenant.

Response example

{
  "tenant_id": "acme-corp",
  "period": "2026-01",
  "cus_used": 12450.5,
  "monthly_quota": 500000,
  "utilization": 0.0249,
  "breakdown": {
    "put": 5000.0,
    "get": 250.0,
    "query_topk": 3200.0,
    "serve": 2000.0,
    "search": 1500.0,
    "delete": 500.5
  }
}

---

Quota enforcement

When a tenant exceeds their monthly CU quota:

1. Requests return 503 Service Unavailable with:

   {"detail": "Monthly CU quota exhausted for tenant 'acme-corp'"}
   

2. The tenant can upgrade their tier immediately via the Console or API.
3. Enterprise tenants can configure soft limits with overage billing.

---

Stripe integration

Managed deployments use Stripe for subscription management:

• Checkout: POST /gate/checkout creates a Stripe Checkout session.
• Provisioning: On successful payment, a webhook (POST /webhooks/stripe) automatically provisions the tenant with the corresponding tier.
• Upgrades/downgrades: Use POST /gate/onboard/update-tier or let the customer manage via Stripe Customer Portal.

Required environment variables:

Variable	Description
HX_GATE_STRIPE_SECRET_KEY	Stripe secret key (sk_live_... or sk_test_...)
HX_GATE_STRIPE_PUBLISHABLE_KEY	Stripe publishable key (pk_live_...)
HX_GATE_STRIPE_WEBHOOK_SECRET	Webhook signing secret (whsec_...)
HX_GATE_STRIPE_PRICE_BUILDER	Stripe Price ID for Builder tier
HX_GATE_STRIPE_PRICE_PRO	Stripe Price ID for Pro tier
HX_GATE_BILLING_ENABLED	Set to true to enforce billing

---

Billing period

• CU counters reset on the 1st of each calendar month (UTC).
• Overage charges are calculated at the end of the billing period.
• Unused CUs do not roll over.