FluidVoice Releases

This release focuses on fixing PII recognizers across multiple countries, adding batch deanonymization and recognizer-level thresholds, and migrating the CI/CD pipeline from Poetry to uv while adding Python 3.14 compatibility.

Jul 22, 2026

Improves pasting reliability in terminal-based apps and fixes audio crashes during device changes, while making the experimental faster recording path opt-in.

Jul 21, 2026

This release references the official Tailscale changelog for details, as no specific changes are listed in the provided text.

Jul 20, 2026

Resolves a crash in the Fluid-1 MLX backend on macOS 15 by bundling the required Swift compatibility runtime.

Jul 14, 2026

Significantly boosts Fluid-1 speed on Apple Silicon, adds Custom Dictionary suggestions, and restores all Whisper models while making spoken punctuation formatting optional.

Jul 14, 2026

A critical hotfix release that fixes a boot-crash in v3.8.47 (missing artifacts) and delivers features like the Langfuse plugin, combo budget fallback policies, and sidebar quick-filtering.

Jul 13, 2026

The original release containing features such as Skill Collector CLI detection, ClinePass dual-auth, external IdP support for Kiro, and combo context requirements configuration.

Jul 13, 2026

A server-only hotfix addressing specific issues where PostgreSQL connections fail to load user data.

Jul 8, 2026

A quick hotfix release containing several bug fixes and the addition of a new settings menu page to display licenses for third-party dependencies.

Jul 7, 2026

Expands Fluid Intelligence capabilities to handle longer text and introduces voice-friendly slash commands, app-aware mentions, and direct punctuation-to-symbol conversion.

Jul 6, 2026

A major update introducing Claude Code plugin-marketplace support, confidence-weighted voting, and Cursor CLI provider support, along with project-level configuration overrides.

Jul 4, 2026

This release introduces desktop hardware token signing and significant UI improvements, including the reintroduction of shared signing and enhanced side menus for comments and attachments.

Jun 30, 2026

The release introduces new recognizers for German, Swedish, and Canadian PII, adds a parameter to control whitespace merging during anonymization, and improves the NLP engine efficiency and security configurations.

Jun 28, 2026

Enhances Custom Dictionary management with menu access and JSON import/export, improves dictation shortcut support, and makes model downloads safer with corrupt-cache detection.

Jun 28, 2026

This release focuses on desktop performance improvements, security fixes, and the cleanup of unused Docker dependencies. It includes significant enhancements such as an upgrade to TypeScript 6 and the removal of ffmpeg from Docker images, alongside various bug fixes for desktop users and team management.

Jun 24, 2026

Introduces Fluid Intelligence, a local on-device AI runtime for smart dictation enhancement, refreshes the onboarding flow, adds local audio history, and introduces adaptive theming.

Jun 22, 2026

This release introduces NVIDIA Nemotron 3.5 speech models and addresses critical startup reliability and Keychain persistence issues.

Jun 4, 2026

NVIDIA has released Cosmos 3, a next-generation open omnimodal world foundation model for Physical AI. It unifies language, images, video, audio, and actions within a single architecture to enable developers to build agents capable of understanding, reasoning, simulating, and acting in the physical world.

Jun 1, 2026

Adds hotkey activation modes, experimental dictation speeds, and app-specific AI enhancement prompts to improve workflow flexibility.

May 29, 2026

A maintenance release addressing Owner-Card readiness for the dream-proxy service to eliminate the need for manual dashboard-api restarts.

May 26, 2026