Skip to content
启枢智能KRoute
Pre-launchKRoute is pre-launch. No product is commercially available yet and we are not charging for anything. Prices, latencies and traffic shown on this site are editable sample values or illustrative animations, not quotes or live data.
KROUTE · AI INFRASTRUCTURE

The foundational hub that connects AI to everything

Compute, model APIs and training data are the three things every AI project needs. KRoute is building one entry point for all of them: call many models with one key, rent GPUs on demand, and source training corpora with clear provenance. For organizations whose data cannot leave the premises, an on-prem gateway is planned.

WHERE WE ARE

Pre-launch, stated plainly

KRoute is pre-launch. No product is commercially available yet and we are not charging for anything. Prices, latencies and traffic shown on this site are editable sample values or illustrative animations, not quotes or live data.

6 product lines

Compute, Router, Serving, Gateway, Corpora and Enablement. All 20 capabilities are currently marked “planned”. None is presented as generally available.

4 tools you can use today

VRAM estimator, rent-vs-buy calculator, routing simulator and corpus health check. They run entirely in your browser and upload nothing. The tool interfaces are in Chinese.

A public API design draft

The router follows the widely used chat-completions request format and adds routing options. Each response states which upstream actually served it.

WHAT WE ARE BUILDING

Six product lines, one hub

LineWhat it is forMore (Chinese)
ComputeOn-demand GPU container instances with images, data volumes, job queues and lab workspaces. Designed for per-second billing.算力租用
RouterOne key and one API for many models. Route by price, latency or a balance of both, with automatic failover and unified usage records.模型路由
ServingDeploy open-source or your own models as dedicated or scale-to-zero inference services.模型托管
GatewayAn on-prem model gateway: a single audited egress point with quotas and redaction. Logs stay on your own machines.私有化网关
CorporaTraining data for large models: pre-training corpora, instruction and dialogue data, and domain-specific sets, each with a provenance and license record.训练语料
EnablementSupporting services for AI research and deployment: evaluation, fine-tuning pipelines, experiment tracking and go-live support.研究支撑
WHY “HUB”

A heavy door turns on a single pivot

The Chinese name 启枢 uses the character 枢, the pivot of a door. Models, compute and data are the heavy door. We want to be the pivot that lets it turn: quiet, in one place, and reliable.

Register interest

Tell us three things by email: what kind of user you are, your rough monthly usage or the GPU type you need, and when you would like to start. We will contact registrants in the order received when a private beta opens.

For technical consulting, support services and purchasing, contact 2396081899@qq.com (email) or 19238427076 (WeChat).