A text-focused model that keeps its attention squarely on language tasks — no image uploads, no audio, just words in and words out. This constraint gives it a clean, predictable interface, though it means multimodal workflows require a different tool. Expect precise, well-structured prose and strong reasoning across written tasks.