FunASR Releases

Released a new CLI version featuring anti-pattern rules, context root configuration in config files, and scope-specific rule waivers.

Jul 22, 2026

Launched Skill 4.0.0 with dice-based direction selection, challenger world reviews, automatic job detection, and browser-based decision interfaces.

Jul 22, 2026

This release focuses on documentation improvements regarding Windows CUDA support and a bug fix for the server component handling OpenAI transcription responses.

Jul 19, 2026

This release introduces a performance optimization by implementing lazy loading for top-level AutoModel exports, which helps reduce initial memory usage and improves import speeds.

Jul 19, 2026

The v1.3.20 release focuses primarily on documentation updates, adding new community integrations and refreshing existing ones across various projects like Optimum Intel and SGLang Omni. It also includes updates to discovery lists and external growth trackers to reflect the latest ecosystem developments.

Jul 19, 2026

This release focuses on documentation updates and announcements, including clarifications for CLI subtitle segmentation, integration with AudioNotes, and guidance for realtime session statistics.

Jul 19, 2026

A patch release addressing a regression in CLI subtitle generation (SRT/TSV output). Fixes include improved timestamp formatting, punctuation model loading, and safe handling of malformed timestamps.

Jul 19, 2026

This release introduces llama.cpp backend support and Windows CUDA runtime assets, along with new logging capabilities for realtime WebSocket sessions.

Jul 18, 2026

Releases prebuilt self-contained binaries for the FunASR llama.cpp runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano with built-in FSMN-VAD, including specific assets for CPU (x64/AVX2) and Windows CUDA.

Jul 18, 2026

FunASR 1.3.16 makes the Fun-ASR-Nano realtime WebSocket service directly installable from PyPI. This release introduces the `funasr-realtime-server` command and adds support for streaming multiple utterances with monotonic absolute timestamps.

Jul 17, 2026

FunASR v1.3.15 focuses on production fixes to improve streaming reliability, fix Fun-ASR-Nano compatibility, and refine CLI and text processing behaviors.

Jul 17, 2026

This release introduces prebuilt self-contained binaries for the FunASR llama.cpp runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano models with built-in FSMN-VAD. It eliminates the need for a Python runtime or local build by providing easy-to-use command-line tools. Users can download quantized models via a helper script and choose between standard x64 or optimized AVX2 assets for CPU compatibility and throughput.

Jul 15, 2026

Introduced tabs integration, meter usage migration features, wallet bonus credit support, and event rejection webhooks.

Jul 14, 2026

This release focuses on documentation updates and bug fixes, highlighting the Fun-ASR-Nano model as the flagship option and improving benchmark visibility.

Jul 13, 2026

Introduces prebuilt self-contained binaries for the llama.cpp runtime, enabling on-device ASR for SenseVoice, Paraformer, and Fun-ASR-Nano without requiring Python or build steps.

Jul 13, 2026

A patch release that fixes ARM and MIPS build issues introduced in the previous version by downgrading the dd-trace-go dependency and adds improvements to forwarding and secondary zone support.

Jul 10, 2026

Enhances DNS transport security and operational reliability with safer DoH/DoH3 handling, improved forwarding configuration, and better dnstap support.

Jul 10, 2026

Fixed Swagger documentation issues.

Jul 9, 2026

Addressed security alerts, refactored configuration, and added a ClickHouse migration runner and caching improvements.

Jul 9, 2026

Patched a gRPC CVE, refactored configuration, and added repo caching and Redis locks for deduplication.

Jul 9, 2026