Skip to main content

🧚 Telegram, cloud delegation in TF Code, and a new wave of models

Released on September 28, 2026

Your agents can now join you on Telegram, and your local TF Code agents can delegate work to cloud agents or escalate to a stronger model mid-session! A fresh wave of models from Anthropic, Google, OpenAI and xAI lands across chat and TF Code — and we have extended support for Google Vertex and Anthropic with the latest models — alongside per-model benchmarks to help you pick with confidence, sharper billing and consumption tracking, and a brand-new website. 🚀

🔌 Telegram Channel​

Talk to your agents on Telegram​

ToothFairyAI now connects to Telegram, so you can reach your agents in the messaging app you already use every day. Send a message, and your agent answers right there — with the same tools, knowledge and capabilities it has everywhere else.

Key Features:

  • Chat anywhere - your agents reply directly in Telegram, alongside your existing channels
  • Full agent power - the same agents, tools and knowledge you use in the app, now in your pocket
  • Quick setup - connect a Telegram bot to your workspace and start chatting in minutes

🎯 Cloud Delegation in TF Code​

Local agents that call in the cloud​

Your local TF Code agents can now hand work to cloud agents. When a task needs more memory than your machine can offer — or when you have enabled the perk and explicitly ask for work to run on the cloud — your local agent can delegate it to a cloud agent while it keeps working. And when a step needs a stronger model, your agent can escalate mid-task and the results flow straight back into your session. It works hand in hand with plan mode, so a carefully planned session can spread work where it runs best.

Key Features:

  • Delegate on cloud - hand a task to a cloud agent when RAM runs short, or when you explicitly ask for it
  • Escalate mid-task - when a step needs more power, escalate it to a stronger model and carry on
  • Plan-aware - plan mode and the escalation kernel work together, so work moves where it runs best
  • Visible handoffs - see when and where work is handed off, so nothing happens behind your back

Note: Cloud delegation and escalation are a perk — they must be enabled on the agent, and agents do not have them by default.

🤖 New Models​

A new wave across Claude, Gemini, GPT and Grok​

A fresh wave of model generations is now available across chat and TF Code, including Claude Opus 5.5, Claude Sonnet 4.6, Gemini 3.8 Flash, Gemini 3.1 Pro Preview, GPT-6 and Grok 4.7. On top of that, we have extended support for Google Vertex and Anthropic with the latest models, so you can run them the way that suits you best.

Key Features:

  • New generations everywhere - the latest Claude, Gemini, GPT and Grok models across chat and TF Code
  • Extended Anthropic support - the latest Anthropic models, ready for your workloads
  • Extended Google Vertex support - tap into the latest Gemini models on Google Vertex
  • Fresh capabilities - updated model capability data keeps pickers, templates and pricing in step with each new model

🧠 Model Quality & Benchmarks​

Pick models with confidence​

Choosing the right model just got easier. Per-model benchmark data is now available when you browse models, so you can compare quality at a glance instead of guessing — and the platform keeps model capabilities and information in sync automatically as new models arrive.

Key Features:

  • Benchmarks at a glance - see quality indicators for each model when choosing where your work runs
  • Always up to date - model capabilities stay in sync automatically, with no manual refreshing
  • Better defaults - chat templates and model listings stay aligned with the latest models

💡 Billing & Consumption​

Sharper tracking, visible savings​

Billing and consumption tracking get more accurate, and your dashboard now shows the savings from cached tokens — so you can see exactly where caching is keeping your costs down.

Key Features:

  • Cache savings on your dashboard - see what cached tokens are saving you in the costing view
  • Reliable billing data - billing records stay in sync with consumption as it happens
  • Bigger workspaces - roomier sub-workspace limits for growing teams

🛠️ Improvements & Fixes​

A refreshed website​

We have launched a new toothfairyai.com — with dedicated models, developers and how it works pages, a built-in AI assistant ready to answer your questions, fresh comparison guides and much better SEO, so finding what you need is easier than ever.

TF Code browser polish​

The integrated browser in TF Code gets a more capable, smoother panel in the web app — better rendering, clearer feedback and a nicer experience while your agent inspects and fixes your pages.

Mobile app fixes​

  • Users management - smoother, more reliable user management in the admin screens
  • General polish - small fixes across the app for a tidier experience

📋 Summary​

This update includes:

  • Telegram channel - reach your agents in the messaging app you already use
  • Cloud delegation in TF Code - delegate work to cloud agents or escalate to a stronger model, a per-agent perk that is plan-aware and visible
  • New models - a new wave across Claude, Gemini, GPT and Grok, with extended Google Vertex and Anthropic support
  • Model quality & benchmarks - per-model benchmarks and auto-synced capabilities when picking models
  • Billing & consumption - cache savings on your dashboard, reliable billing data and bigger workspaces
  • New website - refreshed marketing site with models, developers and how-it-works pages plus an AI assistant
  • Polish - a better browser panel in TF Code and mobile app fixes

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! 🧚‍♀️✨

v0.706.0