Gaiのバージョンを、2024年12月22日から最新の1.2.8まで8件掲載しています。いずれも提供元の公式インストーラーにリンクしています。
2年間で8バージョン・年に約4.5回
| バージョン | 日付 | サイズ | ファイル |
|---|---|---|---|
| 1.2.8最新版 | 2026年10月9日 | — | 64ビット EXE（Inno Setup） リリースノート- Updated text model support: Gemini Flash (2.5, 3.1, 3.5–3.8) and Flash Lite (2.5, 3.1, 3.5).
- Updated image model support: Gemini 3.1 Flash Image and Nano Banana 2.1.
- Free tier: Flash Lite text models only, per Google's policy effective October 9, 2026.
- Centralized network connectivity checks for route changes and app lifecycle events.
- Fixed the missing message input field on Android multi-turn chat screens.
- Improved context compression: snapshots now keep key details more reliably and match the app language.
- Dynamic Thinking now uses `thinking_level: medium` instead of adaptive `thinking_budget`.
- Removed custom `temperature`, `top_p`, and `top_k` parameters |
| 1.2.7 | 2026年5月25日 | 15.5 MB | 64ビット EXE（Inno Setup） 15.5 MB リリースノート- upgrade to Gemini 3.1 Flash Lite and Gemini 3.5 Flash.
- optimize Gemini 3 series: tuned default settings for faster inference.
- home input: added 'chat' and 'new' shortcuts for Short-term Contextual Memory management.
- Contextual Compression: drastically reduce token usage and latency by distilling backgrounds into logic essence, eliminating AI reasoning burden for near-instant, precise responses.
- fixed: file upload for Windows content caching.
- in-app help: added compression guide and draft recovery FAQ |
| 1.2.6 | 2026年3月8日 | — | 64ビット EXE（Inno Setup） リリースノート- support Gemini 3.1 Flash Lite preview.
- support Gemini 3.1 Flash Image preview (Nano Banana 2).
- home input: support short-term contextual memory.
- in-app help: fixed and verified all broken links.
- in-app help: learn how to make AI work more cost-effectively and efficiently for you.
- more models for paid tiers: access gemini pro and multimodal (image) models.
- context caching for paid tiers: reuse data to cut costs and boost speed.
- detailed token breakdown: input/output, caching, and multimodal metrics.
- prompt attachments: increased inline file limit to 50MB.
- optimized UI loading: added a 90s auto-timeout.
- fixed: image gen for usage mode: local.
- fixed: can launch url in markdown content. |
| 1.2.4 | 2025年12月29日 | — | 64ビット EXE（Inno Setup） リリースノート- support Gemini 3.0 Flash preview.
- support Gemini 2.5 Flash image generation.
- support Imagen 4.0 generate (usagemode: serveonly).
- support uploading large files and storing them for 48 hours.
- support secure tunneling for geo-blocking bypass (usagemode: forwardonly).
- remove Gemini 2.0 models are now shut down. |
| 1.2.1 | 2025年7月27日 | — | 64ビット EXE（Inno Setup） リリースノート- upgrade model imagen-3.0-generate-001 to 002.
- support Gemini 2.5 Flash dynamic thinking.
- support Gemini 2.0 Flash preview image generation.
- http job, new option, sendmail on task failure only.
- fix bug http job auto-start on app launch.
- fix bug history attachment save to local direcotry.
- fix bug imagen save to local file. |
| 1.2.0 | 2025年6月25日 | — | 64ビット EXE（Inno Setup） リリースノート- support Gemini 2.5 Flash dynamic thinking.
- support Gemini 2.0 Flash preview image generation.
- support Gemini 2.0 Flash, Multimodal inputs, Text output.
- automate generative AI's scheduled tasks with python.
- on interruption, show the last input for easy chat retry.
- automatically delete logs after a set period to free up space and improve performance.
- setting - Reset Trial - fix trial errors and restart on Home. |
| 1.1.1 | 2025年2月20日 | — | 64ビット EXE（Inno Setup） リリースノート- support Gemini 2.0 Flash, Multimodal inputs, Text output.
- automate generative AI's scheduled tasks with python.
- on interruption, show the last input for easy chat retry.
- automatically delete logs after a set period to free up space and improve performance.
- setting - Reset Trial - fix trial errors and restart on Home. |
| 1.0.0 | 2024年12月22日 | — | 64ビット EXE（Inno Setup） |
