deep-jade•6mo ago
@everyone **DEEPSEEK AND LLAMA NOW ADDED
@everyone DEEPSEEK AND LLAMA NOW ADDED!
Both of these are via Groq -> So they're extra speedy 🏃♂️
A few other recent updates
1. We added a new 'voice widget' that lets you start a call right from the browser!
2. We updated the functions editor to have code snippets you can inject + new starter templates
3. MODEL FALLBACK now is available on the Teams plan (choose a model to fall back too if OpenAI or Anthropic is having an outage)
4. Proactive Messages now supported on the new webchat (and many bug fixes)
7 Replies
fascinating-indigo•6mo ago
More details on model fallback -> if there is an outage from a core provider you can choose the model to auto-fall back too.
This has already been used by a few of our enterprise customers and saved them from the Anthropic outage last week! So we're opening it up to customers on the teams plan aswell!
fascinating-indigo•6mo ago
We're also working on getting Claude 3.7 in with some cool new settings 👀
unwilling-turquoise•6mo ago
It seems that Deepseek's API is unstable, how is it managed?
we're using standard DeepSeek model - I'd reco adding a model fallback in your settings so that if it fails, you can fallback to another provider @Romain
Is the proactive messages feature documented somewhere? Curious what that is
Asked and then immediately found it, nvm
xenial-black•6mo ago
Is this live, as I dont see the "voice widget" feature
fascinating-indigo•6mo ago
@Romain this is via Groq. They host their own version of the model.
@Valeri Sabev It is live! Screenshot above on how to access
@hskrasek yes - docs here https://docs.voiceflow.com/docs/embed-customize-styling#proactive-text-message-bubbles
