Changelog

This page records user-facing changes to features, pages, and the overall experience.

2026-10-13

• Added claude-opus-5-5-kiro model as a Claude Code exclusive with special pricing.
• Kiro pricing: Input $1.20/1M tokens, Output $6.00/1M tokens, more cost-effective than the standard version.
• Ideal for complex code generation, multi-step reasoning, large-context processing, and more.
• Supported in Claude Code CLI, API, and extensions. See the Claude Setup Guide for details.

2026-09-30

• Added gpt-6.1-sol at the same price as gpt-6-sol, including cache and long-context pricing.

2026-09-23

• Added gpt-6-sol and gpt-6-luna for complex and fast everyday tasks; both are billed at 15% of the official price.
• Updated the service support page with recent maintenance updates, announcement links, and contact details.
• Report model access issues via email, WeChat, or Telegram and follow progress through service announcements.

2026-09-12

• Added gemini-3.8-flash; configure an upstream model mapping to use it.

2026-09-05

• Added gpt-6-astra as the new-generation flagship model, billed at 15% of the official price.
• Astra short-context pricing is $1.50 per 1M input tokens and $7.50 per 1M output tokens; cache and long-context rates follow the official multipliers.
• API docs, the Codex setup script, and the managed extension model list now support gpt-6-astra.

2026-07-17

• Organization administrators can receive $5 in wallet credit from the campaign page after joining the official Telegram group. Both new and existing users may participate once.
• After linking the campaign Bot, you can receive website notifications through Telegram and adjust your notification preferences at any time with /stop and /subscribe.

2026-07-10

• Three new GPT-5.6 model tiers are now available: gpt-5.6-sol (flagship), gpt-5.6-terra (balanced), and gpt-5.6-luna (lightweight). All are billed at 10% of the official price.
• gpt-5.6-sol supports max reasoning effort, making it suitable for the most complex coding and long-running tasks.
• The default model in the Codex installation script has been changed to gpt-5.6-sol. See Switching Existing Users to GPT-5.6 for instructions.

2026-05-02

• Image generation now supports background tasks. You can leave the page after submitting a task and return later to check its progress and results.
• Image generation can now process multiple tasks simultaneously, making it easier to try different prompts in batches.
• Image generation failures now provide clearer messages, helping you determine whether the issue is related to your balance, model availability, or service instability.
• Image generation tasks now have more time to complete, improving the success rate of complex image requests.
• Added image API integration documentation to help you integrate gpt-image-2 into your own applications.

2026-05-01

• Added an image generation workspace where you can select a Token, dimensions, quality, and reference images to generate images directly from the console.
• gpt-image-2 now supports pay-per-generation billing, with charges deducted from your wallet balance after successful generation.
• Image generation charges now appear in usage records and wallet transactions, making them easier to reconcile.
• The pricing and billing pages have been separated, making it easier to view plans, top-ups, wallet transactions, and order history.
• The Usage page now displays your wallet balance, making it easier to monitor.
• The remaining weekly pass quota now more accurately reflects your current usage status.

2026-04-30

• Launched a one-time 7-day weekly pass for short-term, high-frequency use.
• Weekly pass quotas are now calculated in consecutive 24-hour periods, providing a more accurate remaining quota.
• Improved the payment and weekly pass activation process.
• Status messages are now more accurate when requests are canceled by the client.
• Added more common scenarios to the error message quick-reference documentation.

2026-04-29

• Improved organization search and usage transaction views in the admin console, making it easier for administrators to investigate usage.
• The Usage page now provides a more transparent breakdown of costs.
• Added a top-up overview to the admin console, making it easier to view top-up trends.
• Standardized chart styles across admin pages for easier data analysis.
• Improved account upload deduplication to reduce duplicate account records.

2026-04-28

• Added a configuration guide for Hermes Agent.
• Model billing details are now displayed more clearly, making costs easier to understand.
• Improved the browser extension connection experience.
• Reduced unnecessary admin alerts when the service is busy.

2026-04-27

• Moved order history to a dedicated page for clearer viewing.
• Redesigned the Tokens page to make the Base URI easier to view and copy.
• Token revocation confirmation now uses a dialog to reduce accidental actions.

2026-04-26

• Added support for using ChatGPT models with Claude Code.
• Added order history, making it easier to review subscription and top-up records.
• The console now supports mobile browsing.
• Users who register through a campaign link can receive exclusive bonus credit.
• Referral rewards now use a separate balance and can be transferred to the wallet or submitted for withdrawal.
• Reorganized the Referral Center, where you can now view reward, transfer, and withdrawal records.
• Standardized the styling of transactional emails.

2026-04-25

• Added a public status page for checking service availability.
• Email notifications are now sent when Tokens are created or deleted.
• Balance alerts are now more timely.
• Improved Codex stability.
• Added GitHub and Google sign-in to the login page.
• Added support and status links after login.

2026-04-24

• Added support for GPT-5.5 Codex models.
• Wallet mode now checks quotas directly against the available balance.
• Added a referral rewards program.
• Updated the homepage branding copy to focus on shared ChatGPT accounts.
• Added the /v1/models compatibility endpoint.
• Added support for Google Gemini models.
• Added spending statistics and filters to the Usage page.
• Email notifications are now sent after successful payments.

2026-04-23

• Added support for the GPT image API.
• Improved the clarity of billing entry points and organization lists.
• Added the Base URI to the Tokens page.
• Added a Chrome smart form-filling extension.
• Recent requests and wallet transactions now display amounts with greater precision.
• Improved payment process stability.

2026-04-22

• Improved wallet fallback handling and quota display after registration.
• Improved usage log parsing accuracy and made recent requests the default view.
• Improved streaming output compatibility.
• Improved the accuracy of Codex token usage statistics.
• Improved the Chinese localization of model pricing displays.

2026-04-21

• Added support for subscriptions, on-demand wallet top-ups, and trial credit.
• Improved quota displays to clearly distinguish between wallet, subscription, and trial credit.
• Expanded the Windows Codex configuration script and documentation.
• Improved the top-up and subscription experience.

2026-04-20

• Registration now supports email verification codes, slider CAPTCHA, and blocking of already registered email addresses.
• Added support for requiring invitation codes during registration.
• Redesigned the console with a left-side navigation layout.
• Simplified the workspace concept and improved page usability on large screens.
• Added usage guides for Cursor, VS Code, and Windsurf.
• Split the documentation into Mac, Windows, and Linux scenarios and added code block copying.

2026-04-19

• Launched the homepage, documentation site, login and registration pages, and basic console.
• Added email verification codes, email notifications, and balance alerts.
• Removed the distinction between personal and team workspaces. Each account now has one personal space by default.
• Added support for Claude and ChatGPT integrations.
• Established basic token-based billing with support for wallets, transactions, and billing periods.
• Added a one-click Codex CLI configuration script and basic integration documentation.
• Improved Claude and Codex request compatibility to ensure basic requests work properly.