VectorBayX logo VECTORBAYX Start a Conversation
Products & Pricing

A short catalogue, priced in the open.

Four product lines covering the full path from securing frontier model capacity to operating open-weight models in your own infrastructure. Pricing below is indicative; every engagement closes with a formal quotation.

Product List

What you can buy from us.

VBX-V · Capacity

Video Generation Capacity

Committed inference quota for a frontier-class video-generation model, structured as prepaid, contractual supply for customers operating outside mainland China. Suitable for consumer creative tools, advertising production, and content platforms where generation throughput is on the critical path.

from HK$120,000 per quarter · reserved tier
Model Class
Frontier video-generation model (upstream provider under NDA)
Tiers
Reserved (quarterly) · Committed (annual) · Enterprise (multi-year, custom)
Structure
Prepaid commitment with agreed service levels and allocation management
Territory
Asia-Pacific markets outside mainland China
SLA
Quota availability commitment with priority escalation path
VBX-D · Engineering

Open-Model Deployment

One-time implementation project that takes an open-weight foundation model from weights to a production endpoint in your data centre or cloud environment — including hardware planning, runtime stack, quantisation strategy, and hardening.

HK$350,000 – 1,200,000 one-time · scope-dependent
Model Scope
Major Chinese open-weight foundation models, incl. trillion-parameter-class MoE systems
Deliverables
Production inference cluster, runbooks, monitoring baseline, handover training
Duration
Typically 6–14 weeks, depending on hardware readiness and model scale
Environment
Customer data centre, regional cloud, or hybrid
Warranty
30-day post-handover stabilisation window included
VBX-O · Subscription

Managed Model Operations

Annual SLA-backed operations subscription for GPU inference clusters running open-weight models: monitoring, incident response, capacity planning, version upgrades, and direct liaison with the model developer’s engineering organisation.

from HK$180,000 per cluster · per year
Coverage
Day-2 operations, patching, version upgrades, on-call incident response
Service Levels
Tiered response times defined in the subscription agreement
Vendor Liaison
Direct engineering escalation channel to model developers
Reporting
Monthly operations report: availability, incidents, capacity outlook
Renewal
Annual; designed for multi-year continuity
VBX-P · Engineering

Inference Performance Engagement

A focused optimisation engagement for teams already running open-weight models: throughput, latency, and cost tuning for large mixture-of-experts systems — batching strategy, KV-cache management, and hardware-specific optimisation.

from HK$150,000 per engagement
Scope
Serving-stack profiling, scheduling policy, quantisation review, cost modelling
Duration
Typically 4–8 weeks
Outcome
Measured improvement targets agreed before kickoff
Prerequisite
Existing production or staging inference deployment
Follow-on
Converts naturally into a Managed Model Operations subscription
Documentation

Technical documents, available under NDA.

Detailed technical and commercial documents are shared with qualified customers and partners under NDA. Request access and we will respond within two business days.

DOC-01 · Capacity

Video Capacity Service Overview

Quota structure, tier definitions, service levels, allocation management process, and commercial terms framework.

Request access →
DOC-02 · Architecture

Deployment Reference Architecture

Reference topologies for open-weight model clusters: memory planning, serving stack, networking, and observability baseline.

Request access →
DOC-03 · Operations

SLA & Operations Handbook

Service-level definitions, escalation matrix, incident classes, reporting cadence, and version-upgrade policy.

Request access →
DOC-04 · Compliance

Security & Compliance Brief

Data-handling practices, access control, confidentiality terms, and cross-border commercial structuring notes.

Request access →
Indicative Pricing

Reference pricing, in HKD.

Indicative · Formal quotation on request
Product Reference Price Notes
VBX-V · Video Capacity — Reserved from HK$120,000 / quarter Entry tier for committed quota; volume-dependent.
VBX-V · Video Capacity — Committed Annual, custom Anchor-scale commitments typically reach eight-figure HKD prepaid volumes.
VBX-D · Open-Model Deployment HK$350,000 – 1,200,000 One-time project fee; driven by model scale, cluster size, and environment.
VBX-O · Managed Model Operations from HK$180,000 / cluster / year SLA tier and cluster count affect final pricing.
VBX-P · Inference Performance from HK$150,000 / engagement Scoped against agreed improvement targets.

All figures are indicative reference pricing in Hong Kong dollars, exclusive of applicable taxes, upstream usage fees, and customer-side hardware or cloud costs. Final pricing is established in a formal quotation following a scoping discussion. Capacity products are offered only for use outside mainland China, within the company’s authorised territory.

Next Step

Send us your workload profile — we will return a scoped quotation.

Start a Conversation →