LongCat-Video Releases

Introduces FluidMeet for local recording with live captions, Global Search across the app, and a redesigned UI with clearer navigation. Adds personalized dashboards, customization options for recording overlays, and new speech models (Mini/Pico). Includes fixes for AI enhancement, dictation responsiveness, and audio capture.

Oct 12, 2026

This release resolves a critical bug where the server crashed on older x86 CPUs lacking SSE4.2 support and improves cross-platform development tooling compatibility for Windows.

Oct 12, 2026

This release focuses on bug fixes for CLI, MCP, and confidence handling, alongside the introduction of a new Go foundation for tokenizer and language detection.

Oct 12, 2026

This release improves model routing stability by ensuring retired models stay retired after restart and adds idempotency conflict checks for API retries when generation options change.

Oct 11, 2026

This release adds support for the Kokoro text-to-speech model from Free.ai and removes automated validation checks for LLMTR providers to prevent account suspensions.

Oct 11, 2026

Enhances the serving pipeline for large choices and improves the TypeScript SDK with tournament selection and cached embedder capabilities.

Oct 11, 2026

A security release addressing missing model authorization, unbound render jobs, and legacy media serving. It introduces a significant breaking change requiring configuration updates for server-managed model authorization.

Oct 11, 2026

This release introduces re-ranking and moderation endpoints, a reliability floor for routing, and custom endpoint grouping, along with support for new providers like Inferbase and Plugsky.

Oct 11, 2026

Oct 10, 2026

Expands the Java SDK with batch prediction and tournament methods, while fixing serve composition and reporting Jev's confidence formula.

Oct 10, 2026

This release focuses on enhancing operator capabilities and AI integration. It introduces a new token scope for job dependencies, allows operators to compose full-code apps, and grants AI session access. Additionally, it improves reliability by adding retry logic for AI failures and changes Deno support to full images only.

Oct 9, 2026

Oct 9, 2026

This is a feature-rich release improving monitoring, AI tooling, and workspace administration. It adds an API base URL configuration, Pi templates for AI, and detailed resource metrics. It also enhances the pipeline editor and allows admins to manage user drafts.

Oct 9, 2026

Focuses on Zoom integration improvements, specifically complete transcripts and precise timing. Moves storage to versitygw for Lite and Compose versions, replacing the previous MinIO setup.

Oct 9, 2026

Oct 8, 2026

Primarily a documentation and bug fix release that improves the Java SDK with lifecycle hooks and enhances training workflows.

Oct 8, 2026

This update enhances the coding agents feature by treating memory as a past record and adding a new TraeCode harness. It also improves system reliability by making bank deletion a no-op and running the embedded daemon out-of-process.

Oct 8, 2026

A maintenance release focusing on bug fixes, parser improvements, and dependency updates for the Rust engine.

Oct 8, 2026

This release focuses on UI/UX consolidation and AI model updates. It introduces a unified page header and sidebar for operators and developers. It also adds a new GPT-6.1-sol model and improves data table permissions and raw app editor security.

Oct 8, 2026

Oct 7, 2026