Rise AI Platform Docs

English

简体中文

English

简体中文

Appearance

v2.2 Release Notes ​

Version: 2.2.0 (release-2606.v2.2.x) Description: This page provides the feature baseline for v2.2, compiled from the frontend codebase and product feature list. Differences from the previous version (v2.1) are listed separately in the "Key Changes" section.

Feature Baseline ​

User Interface (AI Compute Development Platform) ​

  • AI Compute Development
    • Compute Services: Container Instances (Web terminal, VS Code/Jupyter, image saving, auto-shutdown/release, compute adjustment, suspend), Inference Services (deployment templates, version history and rollback, OpenWebUI chat), Custom Services.
    • Model Development: Training jobs, fine-tuning jobs, evaluation jobs, with support for distributed training and queue scheduling.
    • Data Management: Datasets, data cleaning, data augmentation, online augmentation.
    • AI Assets: Image service, storage, credentials and keys.
    • Recycle Bin: Recovery and permanent deletion for container instances, training jobs, fine-tuning jobs, and inference services.
  • LLM Services: Model Marketplace, Online Playground (chat/embedding/reranking/image/video/speech), My Calls and Usage.
  • Personal Center: Account (order overview, transaction overview, API quota overview), API Keys, SSH Keys, SFTP configuration, password change, messages and operation logs.

Management Console ​

  • AI Compute Management: Admin workspace, workspaces and projects (member management, resource pool management, quota management), full-platform compute service management, model management (model registry, deployment templates), AI Gateway (API channels, AI routing, API keys, MCP), AI asset management (specifications, image sources, image service, storage service, tags), policy management (training scheduling policies, training queues, priority, resource reclamation).
  • Cluster Management: Resource overview (node management, GPU management), compute configuration (scheduling policies, overcommitment policies, quota management, virtual GPUs, resource pools, GPU error configuration, monitoring metrics, licenses), images, storage, workloads, and other K8s operations capabilities.
  • Billing: Workspace resource overview (resource billing strategies), physical resource overview (allocation-based billing), Token Governance (model billing strategies, API keys, call records), API channel costs.
  • Alert Center: Alert rules, alert management (subscription policies, message templates, noise reduction), on-call management, analytics dashboard.
  • Platform Management: Platform and members (platform users, platform roles), platform configuration (global parameters, third-party login, OAuth applications, password security, application management).

Common Capabilities ​

  • AI Assistant: A floating intelligent assistant available on any page across both consoles, supporting streaming chat, Markdown responses, and page navigation. The icon is draggable.
  • Top Navigation: Multi-cluster switching, message notifications, page search.
  • Themes: Classic theme and custom themes (light/dark/system-follow, custom primary color).

New Features (2.2.x Iteration Additions) ​

  • OIDC Login: A new OIDC login method is available under Platform Configuration > Third-party Login. Admins can integrate with any OIDC Provider (e.g., Okta, Auth0) by providing Client ID / Secret / Discovery URL.
  • AI Gateway Domain & Certificate Management: Two new pages added to AI Gateway — "Domain Management" (/admin/ai/gateway/domain) and "Certificate Management" (/admin/ai/gateway/certificate). Admins can bind custom domains and upload TLS certificates. AI Routes now include a "Path Prefix" field for domain-based routing.
  • Admin Workbench KPI Card Upgrade: API Quota Overview, Token Overview, and License cards upgraded to server-side pagination, eliminating truncation on large datasets.
  • Admin Workbench Default Project & Shared Projects: Admin workbench supports setting a default project; shared project scope expanded to workspace level.
  • License Expiry Reminder: The LICENSE_EXPIRY_REMINDER parameter is now active. The system sends a notification N days before license expiry (default 30 days).
  • Cluster Management Parameter: New CLUSTER_MGMT global parameter to toggle the cluster management module on/off.
  • Token Governance Dashboard Upgrade: The Token Governance overview now includes a "Time to First Token" card and cache hit rate display, with total Tokens and cost split by input / output / cache; Model Call Records now include a cache hit rate trend, latency / TTFT trends, split Average / P95 latency cards, and filtering by Token usage range (closed interval).
  • API Channel Monitoring Upgrade: The channel overview now includes a "Time to First Token" card (with drill-down), and Channel Quality Monitoring adds a TTFT trend chart.
  • My Usage Details Enhancement: "My Calls and Usage" now shows input / output Token split and cache hit rate.
  • Amount Formatting: Amount and Token values across billing and usage pages are now formatted with thousands separators.
  • AI Route Cross-Cluster Mode: AI Routes' "Route Type" adds a cross-cluster mode, allowing a sub-cluster gateway endpoint pool (address:port + weight) for multi-cluster load distribution.
  • Platform UI Internationalization: The platform UI now supports Chinese and English, with all text consolidated into the i18n system for on-demand language switching.
  • Account Security Enhancements: MFA unbinding now requires secondary verification and confirmation; local account login passwords are encrypted in the browser before transmission.
  • Login Experience Optimization: The login page distinguishes "Management Console / Developer Platform" entries with platform identity badges; login failures distinguish "incorrect username or password" from "account disabled".

Key Changes from v2.1 ​

The following are verified changes identified during documentation review by cross-referencing the v2.1 administrator manual with the v2.2 codebase. This is not an exhaustive changelog.

  • Simplified Platform User Creation Form: Creating a user now only requires entering the username, real name, email, phone number, organization structure, and platform role. The initial password is now set via the "Change Password" action in the list row. The creation form no longer includes "Select Cluster" or "Select Tenant and Project" fields; tenant assignment is handled uniformly through AI Compute Management > Workspaces and Projects > Member Management.
  • Updated GPU Allocation Workflow: On the node side, compute cards are allocated to tenants via Cluster Management > Resource Overview > Node Management "Node Configuration" (scenario/vendor/device plugin) and "Allocate Cards". On the tenant side, GPUs are added to various resource pools (unallocated/shared/dedicated/overcommit) in Workspace Details > Resource Pool Management via the "Add GPU" action.
  • License Management by GPU UUID: The license page displays authorization status and expiration dates using GPU UUID lists, with support for one-click UUID copying for license application.
  • Improved Metering and Billing System: Split into three strategy types — workspace resource billing (usage-based), physical resource billing (allocation-based), and model token billing (built-in global strategy + per-model/tenant strategies).

Generated from Rise AI Platform release-2606.v2.2.x

Copyright © 2026 Rise AI Platform