Changelog

GPT-5.3 and GPT-5.4 are now available on Chat Data, bringing improved reasoning, coding quality, and massively expanded context windows.

Highlights

• GPT-5.3: Next-generation chat model with 128K context window
• GPT-5.4: Latest flagship with industry-leading 1M+ token context window
• Improved reasoning and coding across both models

Introducing AI Actions in Workflow - Bring the power of 833+ prebuilt MCP apps directly into your workflow AI Conversation Block!

Key Features

• Granular Control - Select exactly which AI actions each workflow node can use
• Built-in Actions - Google Search, Order Tracking, Appointment UI triggering, and more
• 833+ MCP Apps - Google Calendar, Slack, GitHub, and hundreds more via MCP integrations
• Variable Passing - Chain actions by passing outputs between nodes

Check the Workflow Builder page for more details.

We're excited to announce three powerful new AI models on Chat Data: Claude 4.6 Sonnet, Claude 4.6 Opus, and Gemini 3.1 Pro. These models bring significant improvements in coding, reasoning, and agentic capabilities to your chatbots.

Key highlights:

• Claude 4.6 Sonnet: 79.6% SWE-bench Verified with 1M token context and top-tier coding performance
• Claude 4.6 Opus: Enhanced agentic planning, debugging, and 128K output for complex tasks
• Gemini 3.1 Pro: 77.1% ARC-AGI-2 reasoning with 1M token context at an affordable price

Available now in your Chat Data model selector.

Notion retraining now automatically detects and syncs sub-pages. New sub-pages are added and deleted sub-pages are removed during retrain, keeping your chatbot's knowledge in sync with your Notion workspace.

Highlights

• Automatic sub-page detection during retrain
• New sub-pages added without manual re-import
• Deleted sub-pages removed automatically
• No workflow changes required

PrestaShop product direct training now correctly assigns variant images to their corresponding product variants. Each variant displays its own images instead of sharing a single set across all variants.

Highlights

• Variant-specific image assignment
• Accurate product representation per variant
• No duplicate or mismatched images
• Automatic during direct training