That's a fair read. Tool calling reliability with sub-4B models is genuinely the hardest unsolved problem in on-device AI right now. The inference engine (Meta

That's a fair read. Tool calling reliability with sub-4B models is genuinely the hardest unsolved problem in on-device AI right now. The inference engine (MetalRT) is production-grade, the pipeline architecture is solid, but the models at this size are still the weak link for complex tool routing. Larger model support (where tool calling is much more reliable) is next on the roadmap. Please stay tuned!