pipecat Releases

This release introduces new features such as multi-subject entitlement support, ClickHouse retries, and the ability to expose total available granted amounts. It also includes a comprehensive set of bug fixes related to billing, subscriptions, and validation logic. Additionally, it contains a breaking change that refactors the static entitlement config type to a string.

Jan 16, 2026

Foundational release introducing user turn strategies, user mute strategies, GrokRealtimeLLMService, and HeyGen LiveAvatar support with major deprecations.

Jan 14, 2026

This release introduces 'thinking' support for Anthropic and Google LLMs, new frames for thoughts, and support for Ultravox Realtime. It also adds `UninterruptibleFrame` and Daily PSTN dial-in support.

Dec 17, 2025

This release introduces new charting capabilities including stacked and candlestick charts, refactors the Dock and Scrollbar APIs for better usability, and enhances the Editor and Input components with LSP support and single-line modes. It also adds a new Settings component and various UI styling improvements.

Dec 8, 2025

This release adds Gradium STT/TTS services, new languages for AsyncAI, and Smart Turn V3.1. It refactors text aggregation to return typed objects with metadata.

Dec 5, 2025

This release adds AWS Bedrock Agent support and `AggregatedTextFrame` for better bot output tracking. It introduces `LLMTextProcessor` and enhances ElevenLabs and MiniMax capabilities.

Nov 27, 2025

This release introduces a new configuration option for table headers and resolves dynamic layout issues within tables.

Nov 21, 2025

This release adds AI Coudtics integrated VAD, ElevenLabs Realtime STT, and Hindi support for Rime. It refactors Deepgram STT error handling and deprecates the Gemini TTS API key parameter.

Nov 19, 2025

A minor maintenance release that deprecates the KrispFilter and removes the LivekitFrameSerializer. It includes a fix for double spaces in assistant messages.

Nov 10, 2025

Added support for Sarvam Speech-to-Text service with streaming WebSocket support, improved OpenAI Realtime LLM service with ToolsSchema support, added audio silence handling, broadcast frame method, metrics logging observer, pronunciation dictionaries, external observers, and new sonic-3 languages support.

Nov 7, 2025

Added DeepgramHttpTTSService with lower latency, SpeechmaticsTTSService, image/audio context support for LLMs, new DeepgramFlux STT event handlers, Cartesia Sonic-3 generation config, and expanded universal LLMContext support to OpenAI Realtime and Gemini Live services.

Oct 31, 2025

Added ability to start bot from /start endpoint with Daily transport, universal LLMContext support for AWSNovaSonic, bulbul:v3 model for Sarvam, new configuration options, trickle ICE support, runtime TTS settings updates, and various fixes for pipeline task cancellation and service issues.

Oct 22, 2025

Added KrispVivaFilter audio filter, folder download support from runner, GeminiLiveVertexLLMService for Google Vertex AI, new GeminiLiveLLMService configuration options (thinking, enable_affective_dialog, proactivity), pipeline error event, and OpenAI service_tier parameter for latency control.

Oct 10, 2025

Fixed pipelines freezing when using interruption strategies and blocking processors like STTMuteFilter by reverting a change from v0.0.88.

Oct 8, 2025

Added Nano Banana models for image generation in GoogleLLMService, HumeTTSService for expressive speech synthesis, updated default model, and deprecated PlayHT services due to PlayHT API shutdown.

Oct 7, 2025

Added WebsocketSTTService base class for websocket-based STT services, DeepgramFluxSTTService for real-time speech recognition, RTVI messages for audio levels and system logs, cached tokens in MetricsFrame, and updated Anthropic LLM service default model.

Oct 2, 2025

Added HeyGenTransport for interactive avatar integration, StrandsAgentProcessor for voice agents, ElevenLabsSTTService, peer connection monitor, memory cleanup improvements, new synchronous events, global location support for GoogleVertexLLMService, on_pipeline_finished event, and RTVI send-text event.

Sep 24, 2025

Added AzureSTTService interim transcriptions, voice cloning support for GoogleTTTS, speaking rate control for GoogleTTTS and OpenAITTTS, push_interruption_task_frame_and_wait method for programmatic interruption, universal LLMContext for AWS Bedrock, video streaming for LiveKitTransport, and OpenAI/Azure Realtime services.

Sep 12, 2025

Added DTMF support for LiveKitTransport, universal LLMContext support for Anthropic LLM service, restored DailyTransport's native DTMF support using Daily's send_dtmf() method, and various fixes for AWSBedrockLLMService and OpenAIImageGenService.

Sep 6, 2025

This release introduces a universal LLMContext system for runtime LLM switching, adds an IVR navigation extension, and enhances audio processing with new filters and transport message frames. It also includes support for WhatsApp calls and reorganizes the transport module structure.

Sep 3, 2025