Hi,
is possible to use the AI Designer with an local model? I have the Claude - thing running. But this consumes my tokens from the Pro account in 2-3h.
FrameworX 10.1.4, Local GPU server 64GB VRAM, qwen/qwen3.6-27b
Hi,
The AI Designer is powered by Claude Code, which connects directly to Anthropic’s API. Currently, using local models in that development workflow is not yet supported.
To reduce token consumption during AI Designer sessions, the main lever is the /compact command. Running it periodically compresses the conversation history and makes a significant difference on longer sessions.
That said, FrameworX 10.1 does include a native Local AI feature for runtime use inside solutions (operator panels and server scripts), which runs fully offline via Ollama at no per-call cost. If that is what you are looking for, the setup is documented here: https://docs.tatsoft.com/display/FX/Local+AI