|
|
|
Processed through Paypal No account required. |







| Files | ||||
| File Name | Rating | Downloads | ||
| Thoth v3.2.0 Thoth v3.2.0 👉 A local-first AI assistant focused on privacy, autonomy, and on-device intelligence. Thoth is a free and open source local-first AI assistant designed to keep your models, data, and workflows on your own machine instead of in the cloud. It combines autonomous tool use, long-term memory, voice input and text-to-speech, vision and screen capture, health and habit tracking, workflows and scheduling, messaging channels, and file and data handling into a native desktop app. Thoth is built to give users personal AI sovereignty with local inference, local storage, opt-in online tools, and no subscriptions or telemetry. Features: Local-first AI assistant with on-device inference, storage, and privacy-focused design Autonomous ReAct agent with 20 integrated tools and 45 operations Long-term memory with semantic recall across conversations Voice input and offline text-to-speech with multiple English voice options Vision support for webcam capture, screenshots, and image analysis Habit and health tracking for medications, symptoms, exercise, mood, sleep, and more Workflows and scheduling for reusable multi-step prompt sequences Messaging channel support through Telegram bot and email integration Gmail and Google Calendar integration with permission controls File, chart, and structured data support including PDF, CSV, Excel, JSON, and Plotly charts Native desktop app with first-run setup wizard, system tray support, and model selection Supports curated local models including Qwen, Llama, Gemma, Mistral, DeepSeek, Phi, and more Supported operating systems: Windows 10, Windows 11 macOS 12 or higher Supported languages: Not Found Changes: v3.2.0 03-16-26 Automatic conversation summarization for effectively unlimited conversation length. Complete rewrite of the memory deduplication system to prevent duplicates and incorrect updates. Added deterministic deduplication using exact normalized-subject SQL lookups. Added cross-category deduplication for background extraction. Added source tracking for memories with live and extraction tagging. Added active thread exclusion to avoid race conditions during background extraction. Extended update_memory() to accept optional subject, tags, category, and source arguments. Added duplicate consolidation utility for near-duplicate memories. Added memory IDs ... |
![]() |
276 | Mar 17, 2026 Siddharth Sachar |
|
| Showing rows 1 to 1 of 1 | Showing Page 1 of 1 | 1 |
OlderGeeks.com Copyright (c) 2026