v1.0.1

Changed

• Bumped client template dependencies (@pipecat-ai/*, Next.js 16, Vite 8, ESLint 10, TypeScript 6, and related type packages) across the vanilla-js-vite, react-vite, and react-nextjs templates.

v1.0.0

Added

• Added Deepgram SageMaker STT and TTS services.

• Added OpenAIResponsesLLMService as a new LLM option (WebSocket-based
  Responses API).

• Added MistralTTSService for streaming text-to-speech using Mistral's Voxtral TTS API.

Changed

• Updated for pipecat-ai 1.0.0.

• Bumped minimum Python version from 3.10 to 3.11.

• Bumped pipecatcloud minimum version to 0.4.4.

Removed

• Removed OpenPipeLLMService (removed upstream in pipecat-ai 1.0.0).

v0.3.3

Changed

• Updated the default system instruction to be more voice-conversation-aware.

• Bumped pipecatcloud minimum version to 0.4.3.

v0.3.2

Added

• Added param_defaults field to ServiceDefinition for specifying fallback values in generated os.getenv() calls. This lets the quickstart work out of the box with only API keys — no model or voice env vars required.
  • OpenAILLMService defaults to gpt-4.1.
  • CartesiaTTSService defaults to voice 71a7ad14-091c-4e8e-a314-022ece01c121 (British Reading Lady).

v0.3.1

Added

• Added support for:
  • NovitaLLMService
  • SarvamLLMService
  • SmallestTTSService
  • XAIHttpTTSService

Changed

• Updated for pipecat-ai 0.0.108:
  • Renamed Grok services from grok_llm / grok_realtime to xai_llm / xai_realtime, matching the new pipecat.services.xai module path.
  • All XAI/Grok services now use XAI_API_KEY instead of GROK_API_KEY.

Removed

• Removed SambaNovaSTTService (removed upstream in pipecat-ai 0.0.108).

v0.3.0

Added

• Added pipecat init quickstart command that creates a new project matching the pipecat-quickstart configuration (SmallWebRTC + Deepgram STT + OpenAI LLM + Cartesia TTS with Pipecat Cloud deployment).

Changed

• Removed the deprecated connect() method from the Javascript client code and replaced with startBotAndConnect().

v0.2.4

Fixed

• Fixed PSTN and SIP dial-out templates to pass enable_dialout=True to configure(), matching the new default in pipecat-ai 0.0.106.

v0.2.3

Changed

• Updated the pcc-deploy.toml template and README to prefer using Pipecat Cloud cloud builds.

v0.2.2

Changed

• Updated support and patterns for pipecat-ai 0.0.105:
  • LLM, TTS, and STT services now use the Service.Settings(...) pattern for configuration (e.g. model, voice, system_instruction).
  • system_instruction moves from LLMContext(messages) to each LLM service's Settings block. LLMContext() is now created with no arguments.

Removed

• Removed HathoraSTTService and HathoraTTSService as Hathora has shut down.

v0.2.1

Added

• Added --list-options flag to pc init that outputs all available service options (bot types, transports, STT, LLM, TTS, realtime, video) as JSON. This lets CI scripts and coding agents discover valid values at runtime instead of hardcoding them.