Changelog

Our model lineup keeps growing

• More model choices: MiniMax-H3 Ref2VA is now on the platform, further expanding our video generation lineup.
• More creative possibilities: Combined image, video, and audio references bring greater flexibility to consistent-character short films, video remixes, and branded content.
• Easy to try and integrate: The Experience Center and video generation API both support the new model for online creation and application integration.

Upload reference files directly for video generation

• The video generation Experience Center now includes a file upload entry, letting you add images, videos, and audio as reference materials directly for a smoother multimodal creation workflow.

Support is always within reach

• The platform support entry is now available as a floating launcher, making it easier to open support, share feedback, and get help while browsing.

Invitation campaigns are coming soon

• The platform's invitation capabilities are ready, with invitation sharing and reward tracking experiences coming soon. Stay tuned for more campaign benefits and ways to participate.

More control over generation tasks

• The open API and Dashboard now support canceling and deleting image and video generation tasks, making it easier to stop queued work or clear completed records as needed.

More multimodal intelligence in chat

• The Experience Center now supports image uploads, enabling multimodal conversations with vision-capable models for a more natural and efficient chat experience.

Richer multimodal creation via API

• The platform's open video generation API now supports Ref2VA, using reference images or videos with optional reference audio to expand creation workflows for character consistency, camera motion, and audiovisual timing.

More multimodal flexibility via API

• The open video generation API now supports FL2VA, expanding multimodal integration capabilities and giving you more flexibility across a wider range of use cases.

DeepSeek-V4.1-Flash is now available

• The platform's model lineup now includes DeepSeek-V4.1-Flash, expanding multimodal vision, long-context, and high-concurrency inference capabilities while giving you a more flexible and efficient option for long documents, coding, and visual Q&A.

Stay up to date by email

• The platform now lets you link a personal email address to your account and subscribe to product updates, making important news and account notifications easier to receive.

Smoother WeChat sign-in and account unlinking

• Sign-in and account-unlinking dialogs now share a consistent layout, with improved compatibility for quick sign-in on desktop.

More complete video model information

• Model details now show all supported resolutions, aspect ratios, and dimensions, making configuration options easier to understand.

Improved costs and billing

• Amount calculations are now more precise, API Key information has been added to image and video charges, and bill details and exports have been improved.

Another step forward for AI service compliance

• According to an announcement from the Shanghai Cyberspace Administration, Qianshi Technology's wylon Cloud Token Factory has officially completed its generative AI service filing under filing number Shanghai-WYLON-202607090183.
• This milestone marks continued progress in wylon's AI service compliance efforts, supporting a safer and more trustworthy model service and developer experience.

An easier start for new users

• New step-by-step guidance walks you through model pricing, identity verification, wallet balance checks, and API Key creation for a smoother first-time setup.

Clearer account and invoice management

• The account overview, usage statistics, and usage limits pages have been refreshed to make account tiers, usage data, and resource limits easier to understand.
• Invoice requests, invoice history, and billing profile management have also been improved. Issued invoices are now delivered directly as email attachments for easier retrieval and recordkeeping.

Text-to-image and text-to-video generation

• The API Manual now covers image and video generation, including API calls, task retrieval, and result access.
• The Experience Center now shows recent generation tasks and results, making creative work easier to manage.
• The Model Catalog now includes dedicated image and video model views with clear specifications and pricing.
• Users can access generation models and benefits included with their account tier.

Organization avatar uploads fixed

• Organization avatar uploads now work as expected, making organization profiles more reliable to maintain.