Changelog

GPT-5-chat deprecation

GPT-5-chat is being deprecated and removed from both the Synthflow UI and API. All customers currently using this model will be automatically migrated to its newer replacement. No action is required on your part. This will happen on March the 20th.

• gpt-5-chat → migrated to gpt-5.2

Clarify end_call_reasons placement on create and update assistant

The OpenAPI spec listed end_call_reasons on both the top-level request body (via Agent_model) and the nested agent object. Only agent.end_call_reasons is applied when creating or updating an agent; a top-level end_call_reasons field is ignored and does not return an error.

The duplicate top-level field was removed from the documented create/update request schema (via Agent_model). GET /assistants and GET /assistants/{model_id} responses still document an optional top-level end_call_reasons on each assistant object where the API may return it, with a note that configuration must use agent.end_call_reasons on writes. The end_call_reasons field on the Agent schema states this explicitly. The End-Call Reasons guide was updated with a full request example.

Updated references:

Unified Changelog

All platform updates are now consolidated into a single Changelog tab. You can filter entries by category using the tags: API Changes, Deprecation Notices, and Product Updates.

Voicemail message configuration and voicemail_message_left end call reason

You can now configure your agent to leave a voicemail message when it detects voicemail, instead of hanging up. The new voicemail_message object on the assistant endpoints supports two modes:

When the agent successfully leaves a voicemail, the call status is left_voicemail and the end call reason is voicemail_message_left.

Note: Enabling voicemail_message requires greeting_message_mode to be set to human. Voicemail detection relies on the agent listening for the initial speaker to determine if it is a human or a voicemail machine.

Updated endpoints:

For more details, see the Call Configuration documentation.

Fallback Voice Provider Support

We have updated our Public API to support fallback voice configurations. These changes allow you to specify a secondary voice provider and model to be used in case the primary synthesizer fails.

API Changes

The assistants API has been updated to include fallback voice fields within the agent object for GET, POST, and PUT requests.

Example usage:

1
{
2
  "name": "My Assistant",
3
  "agent": {
4
    "prompt": "You are a helpful assistant.",
5
    "voice_id": "primary_voice_id",
6
    "provider": "elevenlabs",
7
    "fallback_voice_id": "fallback_voice_id",
8
    "fallback_provider": "deepgram",
9
    "fallback_model": "aura-asteria-en"
10
  }
11
}

Updated endpoints:

Updated supported LLM models

The list of available AI models has been updated. New models have been added and GPT-4o has been removed.

New models:

Updated endpoints:

Custom end-call reasons

You can now define custom conditions that trigger automatic call termination using the new end_call_reasons field on the Agent schema. When the AI agent detects one of these conditions during a conversation, it says a brief farewell and ends the call. This works with both simple prompt and Flow Designer agents.

Updated endpoints:

New response fields:

When a call ends due to a custom end-call reason, the post-call webhook response includes:

For more details, see the End-Call Reasons documentation.

New greeting_message_mode and greeting_message_human_talk_timeout fields on Agent

You can now control how your agent starts a conversation using two new fields on the Agent schema.

greeting_message_mode lets you choose between three behaviors:

greeting_message_human_talk_timeout sets the maximum number of seconds the agent waits for the human to speak first when using human mode. If the timeout is reached, the agent starts the conversation automatically.

Updated endpoints:

For more details, see the Greeting Message Behavior documentation.

Subaccount Minute Limit (max_minutes) now available in Subaccount APIs

The max_minutes field is now included in the response for the Get Subaccount and List Subaccounts endpoints. This field indicates the maximum number of minutes allocated to a subaccount, making it easier to monitor usage limits alongside the existing minutes_used field.

Updated endpoints:

Notes:

• max_minutes is an integer representing the configured cap for the subaccount
• The field may be null if no limit has been set

Knowledge Base Sources API

A new endpoint has been added to retrieve the list of sources attached to a knowledge base.

New endpoints:

• List knowledge base sources: Returns a paginated list of sources for a given knowledge base, including each source’s type, name, content_preview, url, and timestamps (created_at, updated_at). Use the limit and offset query parameters to control pagination.