n8n ships a rebuilt AI Agent node with native multi-model tool calling
Tools, memory and LLM calls are now native node types spanning Claude, GPT-4o, Gemini, Mistral, Groq and OpenAI-compatible endpoints, removing the need for sub-workflows.
TKTomislav Krajina
Published 7 Jul 2026, 13:34Updated 29 Jul 2026, 03:291 min read
n8n has shipped a rebuilt AI Agent node with native tool calling that spans Claude, GPT-4o, Gemini 1.5 Pro, Mistral, Groq and any OpenAI-compatible endpoint.
What moved from sub-workflow to native node
- ▸AI tools, memory and LLM calls are now native node types inside standard workflows
- ▸Builders no longer need separate sub-workflows just to wire up agent behavior
- ▸A new spatial Canvas UI lets builders visually group related node clusters
The direction this points
Folding agent primitives directly into the core node set — rather than treating them as a bolt-on layer — signals that n8n now considers AI agents a standard building block of automation, not a specialized add-on most workflows can ignore.
Source: n8n
