Flan-T5 is Google’s instruction-tuned version of T5, its earlier text-to-text transformer, released in October 2022. Rather than changing the base architecture, Google fine-tuned T5 on more than a thousand tasks phrased as instructions, spanning translation, summarization, reasoning, and classification across many languages. That fine-tuning made it far more usable out of the box than plain T5, which needed task-specific tuning to perform well. Google released the family in several sizes, from an 80 million parameter small version up to Flan-T5-XXL at 11 billion parameters, all under the Apache 2.0 license and freely downloadable on Hugging Face. Its 512-token context window is short by current standards, which limits it to shorter documents and prompts. It remains a common baseline for research on instruction tuning and a lightweight option for teams that want an open, self-hosted model for straightforward NLP tasks without the compute cost of a larger decoder-only LLM.