What's New in Version 0.661.0
Released on December 6, 2025
We're thrilled to bring you another powerful update featuring intelligent Plan Mode for agentic tooling, a new model release, enterprise-grade finetuning, our new Python SDK, and expanded Voice API capabilities! 🚀
🎯 Plan Mode for Agentic Tooling
Collaborative Planning Before Execution
When using Agentic Tooling, you can now enable Plan Mode to have your agent plan and refine its approach with you before executing any actions.
How It Works:
- Activate Plan Mode from the toggle at the bottom left of the chat input field
- Your agent will outline its proposed approach before taking action
- Review, refine, and approve the plan collaboratively
- Execute with confidence knowing exactly what steps will be taken
Benefits:
- Transparency - Understand the agent's reasoning before execution
- Control - Approve or modify the approach before any actions are taken
- Collaboration - Work together with your agent to refine complex workflows
- Accuracy - Reduce errors by validating the plan upfront
🤖 DeepSeek 3.2
New Model Available for All Subscriptions
We're excited to release DeepSeek 3.2 across all subscription tiers!
DeepSeek 3.2 brings improved reasoning capabilities and enhanced performance, giving all users access to cutting-edge AI capabilities.
🦙 Llama 4 Maverick Finetuning
Enterprise-Grade Model Tuning
All Enterprise subscriptions can now finetune Llama 4 Maverick with fully managed LoRA serverless infrastructure.
Key Highlights:
- Significantly Enhanced Capabilities - Llama 4 Maverick brings a major upgrade to finetunable model performance
- LoRA Serverless - Efficient finetuning without infrastructure management
- No Cost Increase - Same pricing, dramatically improved model capabilities
- Fully Managed - We handle the infrastructure, you focus on your data
This represents a substantial leap in what's possible with custom-trained models on ToothFairy.
🐍 Python SDK
Integrate ToothFairy Agents in 5 Lines of Code
We've released the official ToothFairyAI Python SDK, making it incredibly easy to integrate any ToothFairy agent into your Python applications.
Key Features:
- Python 3.8+ Compatible - Works with any modern Python environment
- Minimal Setup - Get started with just 5 lines of code
- Full Agent Access - Leverage all your configured agents programmatically
Installation:
pip install toothfairyai
Quick Start:
from toothfairyai import ToothFairyClient
# Initialize the client
client = ToothFairyClient(
api_key="your-api-key",
workspace_id="your-workspace-id"
)
# Send a message to an agent (non-streaming)
response = client.chat.send_to_agent(
message="Hello, how can you help me?",
agent_id="your-agent-id"
)
print(response.agent_response)
📦 Available on PyPI
🎙️ Voice to Voice API & WebSocket Integration
Open Voice Integration for Any Platform
We're expanding Voice to Voice capabilities beyond Twilio with a new WebSocket-based API for integration with any third-party platform.
Current Capabilities:
- WebSocket Integration - Real-time bidirectional audio streaming for seamless voice agent connections
- Low Latency - Optimised for real-time conversations with minimal delay
- Platform Agnostic - Connect voice agents to any video conferencing or web application
- API Documentation - Comprehensive docs available at apidocs.toothfairylab.link
Note: SIP integration is currently in testing to ensure stability over web connections.
📋 Summary
This update includes:
- 1 new planning feature (Plan Mode for Agentic Tooling)
- 1 new model release (DeepSeek 3.2 for all subscriptions)
- 1 enterprise enhancement (Llama 4 Maverick finetuning with LoRA)
- 1 new SDK release (Python SDK for easy integration)
- 1 API expansion (Voice to Voice WebSocket API for third-party integration)
Thank you for using ToothFairy! We're constantly working to improve your experience. If you have any feedback or encounter any issues, please don't hesitate to reach out to our support team at support@toothfairyai.com.
Keep building magic with AI! 🧚♀️✨
v0.661.0