Koda-IDS Security Models Get a Browser Playground on ZeroGPU
A new Hugging Face Space puts several small defensive-security models behind a browser interface, with full-precision and GGUF options.
A security model you can try in a browser
FlameF0X has put netgoat-ai’s Koda-IDS defensive-security models into a Hugging Face browser playground. The Space, created on October 13, lists two ways to run recent Frontier releases: full-precision inference through Transformers and four-bit GGUF inference through llama.cpp. A third, older release is available only through the GGUF path. The Space’s English instructions describe an evidence-first prompt for analyzing security telemetry, forming multiple hypotheses and suggesting proportionate next steps. During AITrending’s review, Hugging Face reported the Space running on ZeroGPU and its public page returned successfully.
The launch is the demo, not the model
The current Koda-IDS Frontier model card dates its release to October 6, so the new event is access through this community-built interface, not a newly trained checkpoint. Its author describes Koda-IDS as a small model fine-tuned from Liquid AI’s LFM2.5-2.6B for defensive incident reasoning. Those intended uses and the app’s backend options come from the creators’ documentation. AITrending confirmed the page and runtime state, but did not complete a live inference test or independently measure detection accuracy. Security teams should treat its output as investigative assistance, not a verified incident verdict.