FunASR Releases

This release focuses on stabilizing training workflows by fixing DeepSpeed initialization and gradient accumulation synchronization issues. It introduces new SRT output support for the llama.cpp runtime and updates deployment documentation for SenseVoice and vLLM. Additionally, the release bundles the llama.cpp runtime v0.2.0 for multiple operating systems.

Aug 14, 2026

Prebuilt, self-contained llama.cpp binaries for Fun-ASR-Nano, SenseVoiceSmall, and Paraformer models with built-in FSMN-VAD, removing the need for Python or local C++ builds.

Aug 11, 2026

A patch release focusing on aligning email design tokens with the platform system and fixing bugs related to provenance registry completeness and execution file storage keys.

Aug 10, 2026

A major feature release introducing Snowflake and Dynatrace integrations, along with workspace pinning, plan mode, and various UI and performance improvements.

Aug 9, 2026

Adds Smartlead integration and HEIC photo preview support, along with multi-provider embeddings and various performance and bug fixes.

Aug 7, 2026

This release focuses on enhancing CAD-style rendering capabilities and vector visualization, alongside performance optimizations for handling large datasets and glTF models.

Aug 4, 2026

This release introduces LoRA fine-tuning support for the Qwen3 LLM, fixes the Hugging Face paraformer-en alias routing, and enhances output capabilities by adding JSONL transcript/timestamp support to ONNX binaries.

Aug 4, 2026

A maintenance release focused on site improvements, server reliability fixes, and VAD enhancements, including the launch of the FunASR deployment product hub.

Jul 27, 2026

A major update for the FlexPrice billing platform featuring marketplace support, extensive wallet management updates, and new payment gateway integrations (Razorpay, Moyasar).

Jul 27, 2026

Updates the FunASR MCP server to improve agent discovery and updates the container runtime from FunASR 1.3.14 to 1.3.29.

Jul 26, 2026

Maintenance release adding documentation for the Dify Marketplace plugin and fixing SenseVoice to return VAD sentence timestamps.

Jul 24, 2026

Fixes realtime WebSocket finalization reliability and improves SenseVoice subtitle segmentation to better align tags, punctuation, and timestamps.

Jul 24, 2026

Focuses on OpenAI-compatible serving reliability, fixing verbose_json language reporting and ensuring Fun-ASR-Nano reuses cached AutoModel instances.

Jul 23, 2026

Prebuilt self-contained binaries for the llama.cpp runtime, adding Vulkan and CUDA support for SenseVoiceSmall graph execution on Linux and Windows.

Jul 23, 2026

The first fully published, installable release of the FunASR Model Context Protocol server, exposing local FunASR/SenseVoice transcription via stdio.

Jul 23, 2026

Maintenance release focusing on honoring the server hub for default Nano models and updating documentation for community integrations.

Jul 22, 2026

Added realtime postprocess hotwords functionality and bumped the FunASR version.

Jul 22, 2026

Enhanced the FunASR server with new CLI options, fixed client audio segment loss, and introduced Vulkan and GGUF HTTP transcription support.

Jul 22, 2026

Released prebuilt self-contained binaries for the llama.cpp runtime, supporting SenseVoice, Paraformer, and Fun-ASR-Nano with Vulkan and CUDA backends.

Jul 22, 2026

Updated documentation to include recording-inbox community integration.

Jul 22, 2026