Changelog

A record of feature, data, and experience improvements released to Leaderboard.cn users.

v0.1.162026-07-18

Social Share Cards for Frontier Rankings

Added shareable ranking-card images to frontier rankings, with dedicated images and link previews for the homepage and every domain page.

The homepage and every frontier domain gained share actions with classic bars, columns, and compact portrait card layouts; the selected layout is remembered.
Share images update with each release and include a revision identifier, so pasted links show a current card for the corresponding domain.
The share popover supports previewing and downloading images and copying image links; images opened in WeChat can be saved with a long press.
Charts now use organization colors and improved baselines to make score differences among leading models easier to read.

v0.1.152026-07-17

Ranking Experience and Kimi Subscriptions

Improved ranking and catalog experiences and added Kimi pricing and membership information for China.

Frontier ranking markers, heatmaps, and responsive reading behavior were refined.
Pagination, tables, and card interactions are now consistent across model, organization, pricing, and other catalogs, with complete loading states.
Kimi K3 API pricing in CNY was corrected, and four Kimi membership tiers were added.
Sidebar state restoration, theme switching, and keyboard focus feedback were improved.

v0.1.142026-07-17

Kimi K3 and Ranking Data Update

Added Kimi K3 and synchronized the latest public evaluations and Arena ranking data.

Added Kimi K3 model information, official pricing, and complete benchmark results.
Refreshed data snapshots for multiple Arena.ai sub-rankings.

v0.1.132026-07-17

Feedback Service Deployment Compatibility

Improved domain access and health-check configuration for the feedback service.

The feedback API now permits the main site and its subdomains by default.
A dedicated health-check path was added while retaining the original path for deployment compatibility.

v0.1.122026-07-17

On-site Feedback and Empty States

Added on-site feedback and improved the changelog and empty-page experiences.

On-site feedback now accepts reports about data errors, missing models, pricing updates, site issues, and feature requests.
Successful submissions return a reference number, and browser diagnostics are attached only with the user's consent.
The changelog now supports pagination, while empty model comparisons and 404 return navigation have been improved.

v0.1.112026-07-16

Model Identity Governance and ERNIE 5.1

Strengthened model identity validation and added official ERNIE 5.1 evaluation data.

A model identity registry and automated audit reduce duplicate models and incorrect alias mappings.
A set of Arena ranking aliases was corrected to point to the proper model identities.
ERNIE 5.1 and benchmark results from Baidu's official release report were added.

v0.1.102026-07-16

Model Comparison and Link Previews

Added shareable side-by-side model comparisons and made link cards reliably stay current.

Model comparison now supports up to four models across specifications, capabilities, pricing, and scores.
Adding a model no longer overwrites the existing selection, and models outside the top ten now retain their scores.
Model, benchmark, and data-source details gained multi-format sharing.
The site's social preview image was updated to avoid showing an outdated ranking.

v0.1.92026-07-16

Pricing Catalog and Content Distribution

Expanded model-price discovery, search-engine coverage, and off-site subscription options.

A model pricing catalog now supports filtering by organization, provider, and billing currency, including tiered pricing.
Sitemap coverage was expanded, and detail pages gained canonical URLs, dedicated sharing metadata, and structured data.
Evaluation-data RSS, social link metadata, and IndexNow release notifications were added.
A user-facing changelog was introduced, with release-log validation added to the publishing workflow.

v0.1.82026-07-16

Organization Subscriptions and Open-Weight Deployment

Added organization subscription plans and broadened deployment guidance for leading open-weight models.

Subscription-plan information was added to the Anthropic, OpenAI, and Zhipu AI organization pages.
Deployment methods and hardware guidance were added for leading models in the local and open-weight frontier rankings.
Multi-tier subscriptions were split into individual plan cards that link directly to official pages.

v0.1.72026-07-15

Arena Data and Detail-page Stability

Updated Arena.ai data and fixed detail-page rendering and pricing-display issues.

Refreshed data snapshots for multiple Arena.ai sub-rankings.
Fixed React key warnings in model pricing lists.
Improved long data-source table headers and the deployment-method selector.

Home