Skip to content

Tora v1.0.6

Choose a tag to compare

View all tags
@jasondeng843 jasondeng843 released this 05 Oct 13:21
v1.0.6
d9a2c06

简体中文 · English

Tora v1.0.6

ToCode 现在可以使用 Tora 内置模型，无需自行填写模型 API Key。

本次更新

  • ToCode 模型选择器新增“内置模型”分组，提供 DeepSeek Flash、Gemini 3.8 Flash、GPT-6.1 Sol，并保留自定义模型和已有对话选择。
  • 支持相应思考档位、图片输入以及 ToCode 原有的终端、文件、MCP 和代理工具；选择内置模型不会改变已有权限模式或授权范围。
  • 普通 ToChat 聊天不再扣额度。ToChat 工作模式与 ToCode 改为共用订阅工作额度，分别按输入、缓存输入和输出折算内部 Credits；界面实时刷新剩余百分比，额度耗尽时停止内置模型工作请求。
  • 新增“设置 → 账号 → 兑换权益”，支持爱发电订单号一次兑换；赞助者名单移入账号设置。左下角账户菜单显示滚动窗口百分比、释放时间和按当前套餐变化的升级入口，Ultra 及 Ultra x 隐藏升级入口。
  • 新增 Plus、Pro、Max 5x、Max 20x、Ultra、Ultra x 月订阅，接入爱发电付款与订单核验。额度采用滚动 5 小时、7 天或 30 天窗口；续费保留用量，不在周一统一清零。本次未开放附加额度购买。
  • 修复桌面 ToChat 运行配置仍固定为 DeepSeek 的问题，选择 Gemini 或 GPT 后按实际所选模型执行。
  • 修复内置模型的自动命名、子代理请求标识和自我复核请求流程，避免辅助调用占用主任务请求或提前结束工作流程。
  • 退出登录后清理内置模型的本地运行凭据；新同步的账户令牌只用于本地服务运行时，不再明文写入本地配置文件。
  • 修复新任务回复过快时，流连接尚未接上导致回复不显示、一直停留在“思考中”的问题。
  • 补齐“内置模型”、共享额度提示和思考档位相关的多语言文案。

使用与下载

  • 在 ToCode 输入框打开模型选择器，选择“内置模型”分组中的模型。
  • GPT 上游令牌当前仍使用 default 分组，不承诺 codex 0.07 价位。账户额度与提供商实际计费规则沿用现有配置。
  • macOS：Tora-1.0.6-mac.dmg，支持 macOS 13+、Apple 芯片与 Intel 芯片，使用 Developer ID 签名并经 Apple 公证。
  • Windows：Tora-1.0.6-win-x64.exe，支持 Windows x64。安装包未签名，可能出现“未知发布者”或 SmartScreen 提示。Bash 工具需要 Git for Windows。
  • 同时提供 macOS ZIP、两平台自动更新文件、blockmap 和 SHA256 校验文件。

English

Tora v1.0.6 opens built-in models to ToCode without requiring a personal model API key.

What's new

  • Redeem an Afdian receipt number in Settings → Account. Sponsor lists now live in account settings. The account menu shows rolling-window percentages, release times and an upgrade appropriate to the current plan; Ultra and Ultra x hide the upgrade entry.

  • Add a Built-in models group to the ToCode model picker, with DeepSeek Flash, Gemini 3.8 Flash and GPT-6.1 Sol. Keep custom models and existing conversation selections.

  • Support model-specific reasoning effort, images and ToCode's existing terminal, file, MCP and agent tools. Existing permissions and workspace grants remain enforced.

  • Ordinary ToChat chat is free of work quotas. ToChat work and ToCode share subscription credits in rolling windows, with separate input, cached-input and output rates. The UI displays remaining percentages. Six monthly plans support Afdian payment verification; extra-credit purchases are not enabled.

  • Fix the desktop ToChat runtime configuration that still forced DeepSeek after selecting Gemini or GPT.

  • Fix automatic naming, child-agent request IDs and review continuations so auxiliary calls do not conflict with the main request or close the work flow prematurely.

  • Clear local built-in runtime credentials on logout. Newly synchronised account tokens remain in runtime memory instead of being written in plaintext to the local config file.

  • Recover fast replies completed before a new session attaches its event stream, avoiding missing replies and a stuck thinking state.

  • Localise the built-in group, shared quota hints and reasoning-effort labels.

Usage and downloads

Select a model from Built-in models in the ToCode composer. The upstream GPT token continues to use the default group; the codex 0.07 tier is not promised.

  • macOS: Tora-1.0.6-mac.dmg, macOS 13+, Apple silicon and Intel, Developer ID signed and Apple notarised.
  • Windows: Tora-1.0.6-win-x64.exe, Windows x64, unsigned. Bash requires Git for Windows.
  • Includes the macOS ZIP, updater manifests, blockmaps and SHA256 checksums.

官网 / Website: https://ohfun.online/

问题反馈 / Issues: https://github.com/CogniticTeam/ToraAI/issues

Assets 12
Loading