Runbox releases TinyLM: a privacy-first micro-LLM for plugin developers

AI · 3 min read

Runbox's TinyLM targets developers of plugins and small apps who require low-footprint conversational intelligence that can run on developer machines. The micro-LLM is optimized for query understanding, slot-filling, and short-context summarization while keeping model size and memory needs minimal.

Privacy and data residency are central to the product: TinyLM can be fine-tuned on private corpora within a secure enclave and deployed on-premises or on customer-managed cloud instances. Runbox also provides a simplified SDK for delta updates and quantized model serving.

While not designed for broad general knowledge tasks, TinyLM fits nicely into design tool plugins that need short conversations and structured outputs—like extracting acceptance criteria from a brief or generating localized copy variants in-app.