A seven-SKU surface for buying, calling, and proving production AI
The public lineup focuses on Model API, Reserved Inference, Dedicated Endpoint, Dedicated Capacity, VaaS, Managed Deployment, and Accelerator Batch. WIN Incubator stays separate in the top navigation.
Public products
7
Seven launch SKUs
Separate entry
WIN
Incubator stays separate from products
One control plane
API
One entry point for models, routing, billing, and VaaS
Public products
Coding Plan now sits inside Reserved Inference, multimodal lives inside Model API, and WIN Incubator stays separate as a program entry point.
Model API
OpenAI-compatible model access with routing, usage, pricing, and request logs.
Reserved Inference
Reserved token bucket, Coding Plan, and predictable lanes for steady usage.
Dedicated Endpoint
Single-tenant model endpoints with version pinning and stable throughput.
Dedicated Capacity
Monthly GPU or cluster capacity with private delivery and capacity planning.
VaaS
Signed usage evidence for request_id, route, cost, and audit exports.
Managed Deployment
BatchIn-managed serving, deployment, and delivery for enterprise teams.
Accelerator Batch
High-throughput batch lanes for non-interactive jobs and verification-ready exports.
WIN Incubator
Stay tuned
WIN stays visible as a separate program entry in the header and is not mixed into the product SKU menu.